Neuromuscular Responses to Incremental

Caffeine Doses: Performance and Side Effects

JESÚS G. PALLARÉS1, VALENTÍN E. FERNÁNDEZ-ELÍAS1, JUAN F. ORTEGA1, GLORIA MUÑOZ2,
JESÚS MUÑOZ-GUERRA2, and RICARDO MORA-RODRÍGUEZ1
1
Exercise Physiology Laboratory at Toledo, University of Castilla-La Mancha, Toledo, SPAIN; and 2Spanish Anti-doping
Agency, Doping Control Laboratory in Madrid, SPAIN

ABSTRACT
PALLARÉS, J. G., V. E. FERNÁNDEZ-ELÍAS, J. F. ORTEGA, G. MUÑOZ, J. MUÑOZ-GUERRA, and R. MORA-RODRÍGUEZ.
Neuromuscular Responses to Incremental Caffeine Doses: Performance and Side Effects. Med. Sci. Sports Exerc., Vol. 45, No. 11,
pp. 2184–2192, 2013. Purpose: The purpose of this study was to determine the oral dose of caffeine needed to increase muscle force and
power output during all-out single multijoint movements. Methods: Thirteen resistance-trained men underwent a battery of muscle
strength and power tests in a randomized, double-blind, crossover design, under four different conditions: (a) placebo ingestion (PLAC)
or with caffeine ingestion at doses of (b) 3 mgIkgj1 body weight (CAFF3mg), (c) 6 mgIkgj1 (CAFF6mg), and (d) 9 mgIkgj1 (CAFF9mg).
The muscle strength and power tests consisted in the measurement of bar displacement velocity and muscle power output during free-
weight full-squat (SQ) and bench press (BP) exercises against four incremental loads (25%, 50%, 75%, and 90% one-repetition maxi-
mum [1RM]). Cycling peak power output was measured using a 4-s inertial load test. Caffeine side effects were evaluated at the end of
each trial and 24 h later. Results: Mean propulsive velocity at light loads (25%–50% 1RM) increased significantly above PLAC for all
caffeine doses (5.4%–8.5%, P = 0.039–0.003). At the medium load (75% 1RM), CAFF3mg did not improve SQ or BP muscle power or
BP velocity. CAFF9mg was needed to enhance BP velocity and SQ power at the heaviest load (90% 1RM) and cycling peak power
output (6.8%–11.7%, P = 0.03–0.05). The CAFF9mg trial drastically increased the frequency of the adverse side effects (15%–62%).
Conclusions: The ergogenic dose of caffeine required to enhance neuromuscular performance during a single all-out contraction depends
on the magnitude of load used. A dose of 3 mgIkgj1 is enough to improve high-velocity muscle actions against low loads, whereas a higher
caffeine dose (9 mgIkgj1) is necessary against high loads, despite the appearance of adverse side effects. Key Words: ERGOGENIC
AIDS, NEUROMUSCULAR EFFECTS, MUSCLE STRENGTH, MUSCLE POWER, LOAD–POWER RELATIONSHIP

T
he ergogenic effect of caffeine (1,3,7-trimetilxanthine) observed differences between studies. In a recent meta-
on endurance performance is well recognized and has analysis, Astorino and Roberson (2) reported caffeine ergo-
been analyzed at length (7). In contrast, it is uncertain genic effects when testing muscle endurance (increased number
if caffeine ingestion is ergogenic during short-term high- of repetitions to failure) but little evidence to sustain caffeine
intensity exercises. One plausible reason for the disagree- ergogenic effects on maximum strength (one-repetition maxi-
ment between studies is the duration/intensity of the effort mum [1RM]). The effects of caffeine on a single forceful
undertaken. Controversial findings have been reported for action should precede the study of several fatiguing repeti-
exercise performance durations 91 min to the point of tions where metabolite accumulation may hinder the ergogenic
muscle failure (1,3,6,16,19,38), efforts lasting G1 min such effect of caffeine. The present study attempts to address the
us the Wingate test and 20-m sprints (6,12,13,17), and single effects of caffeine on single all-out muscle contractions.
maximal isometric (22,25,27,35), isokinetic (4,5,20), or We have recently reported increases in maximum mus-
isoinertial contractions (3,6,29) lasting only a few seconds. cle strength and power output with caffeine ingestion in
Differences in the muscle groups tested (6,14,37), doses of resistance-trained men (29), whereas others found similar
caffeine ranging from 2 to 10 mgIkgj1, and disparity of results in women (14). In contrast, other studies found
samples used that ranged from moderately active subjects to no effect when administering comparable caffeine doses
APPLIED SCIENCES

resistance-trained athletes may have also contributed to the (3–6 mgIkgj1 body weight) in similarly trained subjects
(3,38) or in collegiate football players (40). This uncertainty
is unfortunate because performance in many sports depends
Address for correspondence: Ricardo Mora-Rodrı́guez, Ph.D., Universidad
de Castilla-La Mancha. Avda. Carlos III, s/n. 45071, Toledo, Spain; E-mail: on brief contractions that require a maximum rate of force
[email protected]. development. Even an ergogenic effect of caffeine on mus-
Submitted for publication February 2013. cle power as low as 5% (29) could influence performance in
Accepted for publication May 2013. these short actions. It thus seems relevant to clarify this issue
0195-9131/13/4511-2184/0 because world-class athletes of sports disciplines that require
MEDICINE & SCIENCE IN SPORTS & EXERCISEÒ high muscle strength and power (e.g., track cycling, Olym-
Copyright Ó 2013 by the American College of Sports Medicine pic weightlifting, or volleyball) are among the ones with the
DOI: 10.1249/MSS.0b013e31829a6672 highest caffeine consumption (36).

2184

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
 The minimal caffeine dose needed to enhance endurance strength for the free-weight full-squat (SQ) and bench press
performance was first established in the classical study of (BP) exercises was 112.5 T 12.6 and 121.0 T 22.7 kg, re-
Graham and Spriet (15) (3 mgIkgj1) and later confirmed by spectively, which accounted for 1.47 T 0.16 and 1.58 T 0.19
Kovacs et al. (24) (3.2 mgIkgj1) and other recent studies when normalized per kilogram of body mass. Most of the
(21) (2 mgIkgj1). During prolonged exercise, a delay in subjects were resident in the sports performance center of
central nervous system fatigue could be behind the ergo- the Region of Murcia (Spain). The subjects were informed in
genic effect of caffeine on endurance performance. In con- detail about the experimental procedures and the possible risks
trast, caffeine effects on neuromuscular performance may and benefits of the project. The study complied with the
occur through a different mechanism involving improved Declaration of Helsinki and was approved by the Bioethics
muscle excitation–contraction coupling (27,28). Because caf- Commission of the University of Murcia. Before participa-
feine could be acting through a different mechanism of ac- tion, written informed consent was obtained from each ath-
tion for endurance than for strength-power performance, the lete, and subjects were informed that they could resign from
dose of caffeine required to activate the mechanism could participation at any time. All subjects were light caffeine
also differ. Some studies suggest that a high caffeine dose consumers (e70 mgIdj1 from caffeinated soda or lyophi-
(5–7 mgIkgj1) is needed to elicit performance effects on lized coffee in milk).
isokinetic strength (4,20). To our knowledge, no study has Experimental design. A randomized, double-blind, cross-
addressed the dose of caffeine needed to improve strength over, placebo-controlled experimental design was used, with
and power during isoinertial contractions of large muscle all subjects serving as their own controls. Participants under-
groups. It is possible that the caffeine dose needed to obtain went the same battery of neuromuscular and biochemical as-
an ergogenic effect may depend on the magnitude of the sessments under four different conditions: (a) placebo trial
resistance that the musculature has to overcome. (PLAC) and three doses of caffeine ingestion: (b) 3mgIkgj1
There is limited information regarding the side effects of trial (CAFF3mg), (c) 6mgIkgj1 trial (CAFF6mg), and (d)
the caffeine doses usually ingested for improving sports 9 mgIkgj1 trial (CAFF9mg). Trials were separated by 48 h to
performance (3–9 mgIkgj1). A review suggests that caffeine avoid any possible fatigue and to allow caffeine washout (23).
ingestion of doses higher than 9 mgIkgj1 can lead to ad- All trials began at 8:00 a.m. to control the circadian rhythms
verse effects such as anxiety, restlessness, and headaches, effects (29). Caffeine (Durvitan, Seid, Spain) was provided
which could negatively affect endurance performance (2). In in gelatin capsules to deliver doses of 3, 6, and 9 mgIkgj1
addition, sleep deprivation due to caffeine ingestion could body mass, respectively. The capsules were ingested 60 min
impair athletic performance when the competition extends before the trial to allow peak blood caffeine concentration
over several consecutive days (31). We propose to study the (10) (Fig. 1). In the trial without caffeine ingestion (PLAC
neuromuscular performance effects of caffeine by increas- trial), subjects ingested placebo capsules filled with the same
ing the dose while observing the side effects in the same amount of dextrose to avoid identification. The amount of
subjects. This analysis would identify the oral caffeine dose additional energy provided by the dextrose (È2 kcal) was
that, while increasing neuromuscular performance, would deemed negligible.
not result in undesirable side effects that may undermine Familiarization. All subjects had previously partici-
caffeine’s ergogenic potential. pated in experiments involving all the muscle strength and
Therefore, the purpose of this study was to find the oral power tests performed in this study. Nevertheless, partici-
dose of caffeine that improves the voluntary contraction and pants underwent seven familiarization sessions before the
power of large muscle groups in resistance-trained athletes. start of the experimental trials to avoid the bias of progres-
Different submaximal loads were used to investigate which sive learning. The last familiarization session, performed in
dose of caffeine enhances either slow-velocity high- the morning (8:00 a.m.) of the third day before the beginning
resistance contractions or high-velocity low-resistance con- of the study, included the determination of the individual
tractions. Second, we examined the side effects associated load (kg) corresponding to 25%, 50%, 75%, and 90% of
with a complete range of caffeine doses and their possible 1RM in the BP and SQ exercises for each subject. To carry
implications for the athletes’ performance. We hypothesized out that assessment, the initial load was set at 20 kg for
APPLIED SCIENCES
that a high caffeine dose (96 mgIkgj1) will be needed to all subjects and was increased in 10-kg increments until
obtain an ergogenic effect on slow-velocity high-resistance the attained mean propulsive velocity (MPV) was less than
contractions, whereas lower caffeine doses (G6 mgIkgj1) 0.5 mIsj1 in the BP or less than 0.8 mIsj1 in the SQ because
will be ergogenic in high-velocity low-resistance contractions. those velocities indicate proximity to 1RM (32). Thereafter,
the load was adjusted with smaller increments so that 1RM
could be precisely determined. The heaviest load that each
METHODS subject could properly lift while completing the full range of
Subjects. Thirteen highly resistance-trained men volun- motion was considered to be his 1RM.
teered to participate in this study (age, 21.9 T 2.9 yr; body mass, Experimental protocol. The day before and during the
76.5 T 8.5 kg; height, 172.7 T 5.4 cm; body fat, 12.4% T 2.7%; 7 d that the experiment lasted, the subjects lived at the sports
resistance training experience, 7.1 T 3.5 yr). Their 1RM performance center where they slept and ate all meals. They

CAFFEINE DOSE AND NEUROMUSCULAR RESPONSE Medicine & Science in Sports & Exercised 2185

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
 FIGURE 1—Experimental protocol.

all consumed a diet of 2800–3000 kcalIdj1, composed of 75%, and 90% of 1RM) for upper and lower body muscu-
55% energy intake from carbohydrates, 25% from fat, and lature (BP and SQ). Those step measures allowed a contin-
20% from protein, evenly distributed across three meals uous representation of the load–velocity and load–power
each day (breakfast at 7:00 a.m., lunch at 13:30 p.m., and curves to study the interaction between load and caffeine
dinner at 20:00 p.m.). Subjects refrained from physical ac- dose on neuromuscular performance. Cycling peak power
tivity other than that required by the experimental trials and output (PPO) was assessed next using a nonfatiguing iner-
withdrew from alcohol, tobacco, and any kind of caffeine tial load test of 4-s duration. Subjects remained blinded to
intake 10 d before testing and while the experiment lasted. the results during the whole experiment. Instructions before
The day before the onset of the experiment, height was lifting were standardized and always delivered by the same
measured in the morning to the nearest 0.5 cm during a experimenter.
maximal inhalation using a wall-mounted stadiometer (Seca Upon completion of the test battery (È60 min from the
202; Seca Ltd., Hamburg, Germany). beginning of the neuromuscular assessments), a second
In every trial, upon arrival to the testing facility at urine and blood sample were collected (POST). Then, sub-
6:30 a.m. in a fasted state (PRE), a urine sample (15 mL) jects filled out a questionnaire (QUEST + 0 h) aimed to
was obtained. Urine specimens were measured in duplicate address whether side effects of caffeine were present during
for urine-specific gravity (USG; Uricon-NE, Atago, Japan), the trial. Subjects were then discharged and reminded about
and the rest of the sample was immediately frozen at j20-C their schedule for the next trial. Blood and urine caffeine
for future analysis. Then, the subject’s body weight was concentration was evaluated at the beginning (PRE) and
determined and body water estimated using a four-contact immediately at the end (POST) of each trial. USG and blood
electrode body composition bioimpedance analyzer (Tanita hematocrit were also determined PRE and POST each trial.
TBF-300A; Tanita Corp., Tokyo, Japan) to obtain a percent- Load–velocity and load–power relationships. We
age of body fat and fat-free mass. Following this, tympanic used a graded loading test in a Smith machine (Multipower
temperature (Thermoscan, Braun, Germany) was measured Fitness Line, Peroga, Spain) with a linear encoder and its
in triplicate after the removal of earwax when needed. associated software (T-Force System; Ergotech, Murcia, Spain;
Next, a 5-mL blood sample was withdrawn from an ante- 0.25% accuracy) attached to the bar by a light retractable
cubital vein without stasis. A small portion of the whole blood metal cable. There were two Smith machines, each one
was used to determine hematocrit by triplicate using no- dedicated to a given exercise (SQ or BP). Both encoders
heparinized capillary tubes (70 KL; Hirschmann Laborgerate, were cross validated before the test with agreement of
Germany) and a microcentrifuge (Biocen, Arlesa, Spain). r = 0.999. A detailed description of the validity and reli-
The serum and plasma obtained after centrifugation (3000g) ability data of the dynamic measurement system (ICC = 1.00,
was immediately stored at j70-C. Then, subjects ingested CV = 0.57%) has recently been reported (32). At the indi-
the capsules containing their individualized-randomized caf- vidually determined 25%, 50%, 75%, and 90% of 1RM
APPLIED SCIENCES

feine dose (3, 6, or 9 mgIkgj1) or placebo with 330 mL of a (see Familiarization section), changes in bar displacement
fruit milkshake (168 kcal) and a pastry (456 kcal) that served velocity and power output were measured after the inges-
as a standardized breakfast (total of 624 kcal and 68 g of tion of different caffeine doses. MPV and mean propulsive
carbohydrate). power (MPP) were calculated as the average velocity and
After a standardized warm-up that consisted of 10 min power output values, respectively, measured only during the
of jogging at 10 kmIhj1 and 10 min of static stretches and propulsive phase, defined as that portion of the concentric
joint mobilization exercises, the subjects entered the labo- action during which the acceleration is greater than accel-
ratory to start the neuromuscular test battery assessments eration because of gravity (33). In each trial, three attempts
under a paced schedule (see Fig. 1). These tests consisted of were executed for light (25% RM), two for medium (50%
the measurement of bar displacement velocity and muscle RM), and only one for the heaviest (75% and 90% RM)
power output against four incremental loads (25%, 50%, loads interspersed with 5 min of passive rests. Only the best

2186 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
repetition at each load, according to the criteria of fastest graphy vial containing 900 KL of mobile phase (aqueous
MPV, was considered for subsequent analysis. solution of 0.1% acetic acid). Subsequently, 20 KL of internal
The individual range of movement during the BP and SQ standard working solution (caffeine 13C3 5 KgImLj1) was
exercises was carefully replicated in each trial with the help added and mixed. Ten microliters of the sample was then
of two telescopic bar holders with a precision of T1.0 cm. directly applied to the HPLC-MS system. For blood, 20 KL
In the BP, the bar holders were positioned to allow the bar of the internal standard working solution was added to
to descend to 1 cm above each subject’s chest. In the SQ, the the aliquots of plasma sample (100 KL). The sample was
bar holders were set at each subject’s lowest squat depth vortex mixed for 10 s, then 20 KL of 20% perchloric acid
defined as that position in which the back of the thighs and was added, and the sample was vortex mixed for 10 s and
upper calves made contact with each other. Subjects were centrifuged at 3500 rpm for 10 min. One hundred microliters
instructed to perform the eccentric phase of both exercises of the supernatant was transferred into a liquid chromatogra-
in a slow and controlled manner, to pause for 2 s at the bar phy vial containing 900 KL of mobile phase and mixed. Then
holders, momentarily releasing the weight, and thereafter to the sample was filtered through 0.2 Km of cellulose acetate
perform a purely concentric action pushing back up at max- membrane, 25-mm syringe filters, and 10 KL was then di-
imal intended velocity. The momentary pause imposed be- rectly applied to the HPLC-MS system. To calibrate the
tween the eccentric and the concentric actions was designed system, aqueous solutions of caffeine (ranging from 0.25 to
to minimize the contribution of the stretch-shortening cycle 12 KgImLj1) were used for each batch of samples. The lower
(i.e., rebound effect) and to allow for more reliable and limit for the accurate quantization of these methylxanthines
consistent measurements. Pilot data collected during the was 0.25 KgImLj1.
previous familiarization revealed significant reductions in Side effects evaluation. Immediately after each neu-
the intrasubject coefficient of variation when using the de- romuscular test battery (QUEST + 0 h) and 24 h later
scribed technique compared with a nonstop technique in- (QUEST + 24 h), participants answered a questionnaire. The
volving the stretch-shortening cycle (i.e., 3.5% vs 2.4% for QUEST + 0 h was designed to evaluate the physical fatigue,
BP and 3.7% vs 2.7% for SQ exercise, both P G 0.05). the perceived performance, and the side effects (e.g., urine
Cycling peak power test. Cycling PPO was measured output, gastrointestinal problems, tachycardia, or headache)
using the previously described isoinertial load test (26). In felt by the participants during the neuromuscular test battery.
brief, we measured the power needed to overcome the in- QUEST + 24 h was designed to evaluate physical fatigue
ertial load of a heavy (21.5 kg) cycle-ergometer flywheel and side effects (e.g., sleep quality, gastrointestinal prob-
(Monark-818, Varberg, Sweden). This cycling PPO assess- lems, tachycardia, muscle soreness, or headache) perceived
ment lasts only 4 s; however, a complete power–velocity by participants during the 24 h after the caffeine dose was
spectrum curve is generated using an absolute encoder (ASM ingested. These surveys included eight items on a yes/no
2000 ppr; Unterhaching, Germany; 1000 Hz) connected to the scale and were based on previous publications about side
cycle ergometer flywheel. Subjects sat on the cycle ergometer effects derived from the ingestion of caffeine (8,11).
after the handlebars and saddle had been adjusted to fit their Statistical analysis. The Shapiro–Wilk test was used
individual’s body dimensions. After a 3-min warm-up (100 W to assess normal distribution of data. Pretesting conditions,
at 90 rpm interspersed with two short (2–3 s) bouts of maximal the cycling isoinertial load power test and the caffeine levels
acceleration), subjects performed two maximal sprints inter- data were analyzed using one-way ANOVA for repeated
spersed by 180 s of active recovery (50 W). The test started measures (doses of caffeine). The load–velocity and load–
from a complete stop with the pedal of the dominant leg power relationships were analyzed using two-way (caffeine
placed at 45- from the vertical. The test–retest intraclass cor- dose  load) ANOVA for repeated measures. The Greenhouse–
relation coefficient was 0.85 (0.60–0.95), and the coefficient Geisser adjustment for sphericity was calculated. After a sig-
of variation was 3.9% T 1.3%. The average value of the two nificant F-test, differences among means were identified
PPO sprints was recorded for data analysis. using pairwise comparisons with Bonferroni’s adjustment.
Urine and plasma analysis. Blood samples (5 mL) The significance level was set at P e 0.05. Cohen’s formula
were mixed with ethylenediaminetetraacetic acid in plastic for effect size (ES) was used, and the results were based on
APPLIED SCIENCES
tubes and plasma immediately separated by centrifugation the following criteria: 90.70 large effect, 0.30–0.69 moderate
(MPW-350R; MedInstruments, Poland). The plasma samples effect, and e0.30 small effect (9). Reported side effects in
were stored at j80-C for future analysis. At a later date, urine the questionnaires were not normally distributed, and a non-
and plasma samples were analyzed for caffeine concentra- parametric statistical technique was used.
tions and related metabolites using an Agilent Technologies
HPLC 1200 system (Santa Clara, CA) coupled to a triple
RESULTS
quadrupole/ion trap mass spectrometer (MS; API 4000, Q TRAP,
AB SCIEX, Framingham, MA US). Methylxanthine internal Pretesting conditions, hydration status. Before the
standards were purchased from Cerilliant (Round Rock, TX). four experimental trials (PRE), body mass (range between
Aliquots of urine sample (100 KL) were filtered (VWR, 76.4 T 8.5 and 76.9 T 8.2 kg) and body bioimpedance (range
Barcelona, Spain) and transferred into a liquid chromato- between 462 T 46 and 475 T 54 6) were not different between

CAFFEINE DOSE AND NEUROMUSCULAR RESPONSE Medicine & Science in Sports & Exercised 2187

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
 TABLE 1. Side effects reported by participants immediately after the conclusion of each neuromuscular test battery (QUEST + 0 h) and 24 h later (QUEST + 24 h).
PLAC CAFF3mg CAFF6mg CAFF9mg
+0 h +24 h +0 h +24 h +0 h +24 h +0 h +24 h
Muscle soreness 15 8 8 8 8 31 15 38
Increased urine output 8 8 15 8 15 23 62 54
Tachycardia and heart palpitations 8 0 15 0 15 0 23 23
Anxiety or nervousness 8 0 8 0 15 0 31 23
Headache 8 0 0 8 8 23 15 38
Gastrointestinal problems 0 8 8 8 8 8 31 38
Insomnia — 0 — 0 — 8 — 54
Increased vigor/activeness 8 0 38 0 46 8 54 23
Perception of performance improvement 8 — 54 — 54 — 62 —
Data are presented as the percentage of prevalence.

trials (PLAC, CAFF3mg, CAFF6mg, and CAFF9mg). No sig- CAFF9mg increased the frequency of all adverse side effects,
nificant differences were detected between treatments in with a frequency of appearance from 23% to 54%. Of note,
PRE testing conditions for tympanic temperature, blood he- 23% of participants reported tachycardia and anxiety or
matocrit, or urine USG. POST tympanic temperature values nervousness, 38% with gastrointestinal problems, and 54%
in all trials (PLAC, CAFF3mg, CAFF6mg, and CAFF9mg) with insomnia or sleep disturbances (Table 1).
were significantly elevated (range of increase = 1.5%–2.2%, Load–velocity and load–power relationship. MPV
P = 0.000–0.041, ES = 1.10–1.70) when compared with attained against the two lower loads (25% and 50% 1RM)
their respective PRE value. No significant differences were in the BP and SQ exercises significantly increased with all
detected for hematocrit or urine USG values between PRE caffeine doses (CAFF3mg, CAFF6mg and CAFF9mg) com-
and POST conditions at any dose, except in the CAFF9mg pared with the placebo treatment (PLAC) (range of in-
trial where hematocrit was significantly higher (from 45.4% T crease = 5.4%–8.5%, P = 0.039–0.000, ES = 0.76–1.28;
3.7% to 47.0% T 3.9%, P = 0.031, ES = 0.42), and urine USG Fig. 2). Similarly, MPV at 75% 1RM was significantly
was significantly lower (from 1.024 T 0.005 to 1.016 T 0.007, increased in all caffeine trials in the BP and SQ exer-
P = 0.003, ES = 1.40) in the POST testing conditions. cises compared with the placebo treatment (PLAC) (range
Caffeine side effects. Immediately after the PLAC
trial, subjects reported a very low frequency of side effects
(0%–8%; QUEST + 0 h). The CAFF3mg and the CAFF6mg
treatments produced very similar side effects, with a limited
increase in the sensations of tachycardia and heart palpita-
tions, self-reported urine output, and gastrointestinal prob-
lems (8% of the subjects) compared with the PLAC trial. On
the other hand, the subject’s perception of performance and
vigor increased five to seven times above PLAC during the
CAFF3mg and CAFF6mg trials (38% and 54% of the sub-
jects, respectively). Finally, the CAFF9mg trial produced a
drastic increase in the reported frequency of side effects
(Table 1). It is particularly relevant that 62% and 31% of
participants reported an increase in the estimates of urine
output and gastrointestinal problems, respectively. The per-
ception of performance and vigor or activeness also rose in
62% and 54% of the participants, respectively (Table 1).
The following morning of each experimental trial
(QUEST + 24 h), very few participants (8%) reported that
APPLIED SCIENCES

PLAC treatment produced residual side effects such as

muscle soreness, increase in the estimates of urine output,
and gastrointestinal problems. The CAFF3mg trial produced
very similar side effects to PLAC, with an additional 8% of
participants reporting a headache. The CAFF6mg trial tended
to increase the frequency of muscle soreness and head-
aches and the estimates of urine output in comparison with
the PLAC and CAFF3mg treatments, although still with a
frequency lower than 31% of the subjects. In addition, FIGURE 2—Dose–response effects of caffeine ingestion on load–
velocity relationship for bench press (A) and full squat (B) exercises. Data
CAFF6mg produced symptoms of increased vigor and sleep are means T SD. *Significant differences (P e 0.05) compared with the
problems, although with a very low incidence (8%). Finally, PLAC trial within each load.

2188 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
of increase = 6.3%–8.9%, P = 0.046–0.014, ES = 0.91–1.21;
Fig. 2). The only exception was for the CAFF3mg trial in the
BP exercise (P = 0.15). Finally, MPV at the heaviest load
(90% 1RM) was significantly enhanced with the CAFF9mg
compared with placebo in both exercises (BP: 13.1%,
P = 0.031, ES = 0.74; SQ: 10.4%, P = 0.046, ES = 1.03).
During SQ at 90% 1RM, a caffeine dose of CAFF6mg
was also enough to increase MPV above PLAC (8.3%,
P = 0.029, ES = 0.90; Fig. 2B).
MPP at the two light loads (25% and 50% 1RM) in the
BP and SQ exercises was significantly increased with all
caffeine doses (CAFF3mg, CAFF6mg and CAFF9mg) com-
pared with PLAC (range of increase = 8.1%–12.0%,
P = 0.022–0.000, ES = 0.36–0.68), except for the CAFF3mg FIGURE 4—Dose–response effects of caffeine ingestion on cycling
treatment in the SQ exercise (P = 0.18–0.09). At 75% 1RM, PPO using a 4-s inertial load test in absolute and fat-free mass nor-
MPP in the BP and SQ was also significantly increased malized values. Data are means T SD. *Significantly different (P e 0.05)
than PLAC.
in the CAFF6mg and CAFF9mg trials compared with PLAC
(range of increase = 8.3%–10.2%, P = 0.037–0.010,
ES = 0.36–0.48). Finally, MPP at 90% 1RM was signifi- 1RM, independently of the caffeine dose ingested (range =
cantly enhanced in the CAFF9mg trial for BP and SQ 501–562 W), and was significant lower at 75% and 90%
(11.7%–15.0%, P = 0.031–0.014, ES = 0.47–0.92) and with 1RM loads (range 267–423 W, P G 0.001; Fig. 3A). The
the CAFF6mg dose for the BP exercise (11.4%, P = 0.021, peak in mean power for SQ occurred at 75% 1RM in all
ES = 0.71) compared with PLAC (Fig. 3A and B). The trials (range = 525–579 W) and was significantly lower
peak in mean power in the BP exercise occurred at 25% at 25% 1RM (range 356–400 W, P G 0.001; Fig. 3B).
Cycling PPO test. No significant differences were de-
tected in cycling PPO in absolute (W) or normalized per
kilogram of FFM (WIkgj1) values between PLAC, CAFF3mg,
and CAFF6mg trials. However, a significantly 7.0% higher
PPO was detected in the CAFF9mg trial (1506 T 225 W) com-
pared with PLAC (1408 T 189 W, P = 0.040, ES = 0.47).
Likewise, a significantly 6.9% higher PPO/FFM was de-
tected in the CAFF9mg trial (22.7 T 1.8 WIkgj1) com-
pared with the PLAC trial (21.2 T 1.4 WIkgj1, P = 0.036,
ES = 0.88; Fig. 4).
Urine and blood analysis. Upon arrival to the labo-
ratory (PRE), plasma (G0.13 T 0.08 KgImLj1) and urine
(G0.12 T 0.15 KgImLj1) caffeine concentrations were neg-
ligible in all subjects, confirming the complete caffeine
washout before trials. At the end of the test battery (i.e.,
POST: 2 h after the caffeine or placebo ingestion), urine
and plasma caffeine concentrations in all caffeine trials
(CAFF3mg, CAFF6mg, and CAFF9mg) were significantly
higher (P G 0.05) than their respective basal values. The
ingestion of the graded caffeine doses produced a signifi-
cant parallel increase in plasma and urine caffeine concen- APPLIED SCIENCES
trations at their respective POST values (P G 0.05; Fig. 5).

DISCUSSION
The main finding of this study is that caffeine signifi-
FIGURE 3—Dose–response effects of caffeine ingestion on load–power
relationship for bench press (A) and full squat (B) exercises. Data are cantly improves movement velocity under all loading con-
means T SD. *Significant difference (P e 0.05) between CAFF3mg and ditions (from 25% to 90% 1RM) in both the upper (BP)
PLAC. †Significant difference (P e 0.05) between CAFF6mg and PLAC. and the lower body (SQ) musculature. The higher the load,
‡Significant difference (P e 0.05) between CAFF9mg and PLAC.
a
Significantly lower (P e 0.05) than the peak power (25% 1RM for BP and thus the longer time available to apply force, the higher
and 75% for full squat) for each exercise. the caffeine dose needed to achieve an ergogenic effect

CAFFEINE DOSE AND NEUROMUSCULAR RESPONSE Medicine & Science in Sports & Exercised 2189

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
 loads and an accurate measure of bar velocity are compara-
ble with those using isokinetic tests in that caffeine is ergogenic
during single muscle actions against several external resistances.
We found that a high dose of caffeine (9 mgIkgj1) was
needed to improve peak power in our highly reproducible
(CV = 3.9%) and sensible (sampling frequency 1000 Hz)
inertial load cycling test (Fig. 4). In this test, subjects have
to overcome the inertial load of a heavy (21.5 kg) cycle-
ergometer flywheel. Thus, during the initial pedal strokes of
the 4-s test, the leg musculature is required to develop a high
percentage of their MVC (26). In agreement with the SQ
FIGURE 5—Dose–response effects of caffeine ingestion on urine and
plasma caffeine concentrations at PRE and POST time points. Data are data (Fig. 3B), only the highest dose of caffeine was ergo-
means T SD. *Significantly different (P e 0.05) when compared with genic against the high inertial load. Our data are similar to
their respective POST value. †Significantly different (P e 0.05) when those of Glaister et al. (13) in that we did not find an effect
compared with their previous caffeine dose trial.
of caffeine on cycling sprint peak power with doses of 3
and 6 mgIkgj1. However, we found an ergogenic effect
(see Fig. 3). We have previously reported a È5% MPV with 9 mgIkgj1, whereas they did not with doses of 8 or
improvement in BP and SQ exercises after ingestion of 10 mgIkgj1. However, Glaister et al. reported a tendency for
3 mgIkgj1 of caffeine against loads of 75% 1RM (29). In the a reduction in the time to peak power at the highest caffeine
present study, we observe improvements between 5.2% and dose. Our tests differed in the length of the sprint; ours being
13.1%, depending on the dose of caffeine ingested and less than half of the duration compared with that of Glaister
the magnitude of the resistance to overcome. Interestingly, et al. (i.e., 4 vs 10 s). It seems that the longer the duration of
the ergogenic effect of caffeine against the lighter loads the sprint, the less likelihood of finding an effect of caffeine
(25%–50% of 1RM) was maximal with a low dose of caf- ingestion. In fact, studies using the regular 30-s Wingate test
feine (3 mgIkgj1). In contrast, higher caffeine doses were to investigate the effects of caffeine are inconclusive with
required to improve performance against higher loads either positive (39) or negative (6,17,38) findings. In sum-
(6 mgIkgj1 for 75% 1RM and 9 mgIkgj1 for 90% 1RM mary, our results using repeated muscle contractions (4 s
load). This seems to suggest that the dose of caffeine re- cycling sprint against an inertial load; Fig. 4) confirm the
commended will depend on the resistance that athletes have results observed during single muscle actions (Fig. 3) in that a
to overcome. As we will discuss latter, administering the high dose of caffeine is needed when the resistance to over-
minimal ergogenic dose would be recommended to avoid come is high.
undesirable adverse side effects. The results of the present study show a greater caffeine
As observed in Figure 2, caffeine increased movement ergogenic effect on the lower compared with the upper
velocity in 20 of the 24 caffeine ingestion trials (83% ef- body musculature, particularly at the higher resistances and
fectiveness). This consistency on the ergogenic effects of caffeine doses (Fig. 2). These results are consistent with
caffeine has no comparison in the literature that examines previous findings by Astorino et al. (1), who found an ergo-
caffeine effects on resistance exercise. Some studies found genic effect of caffeine ingestion on repetitions to failure
no effect of caffeine on maximum strength measured as for leg press, but not for BP exercise. In a recent meta-
1RM (1,3,38) or the number of repetitions to failure against analysis, Warren et al. (37) found more consistent effects of
a submaximal load (1,3,6,16,19,38–40). However, our data caffeine ingestion on improving 1RM knee extension than
are not directly comparable with those studies addressing the that in other muscle groups. In contrast, other researchers
effect of caffeine ingestion using the 1RM test. The repro- found enhanced 1RM strength (6) and number of repeti-
ducibility (CV = 2.9%–5.3% [34]) and accuracy (usually tions to failure (39) in the BP but not in the leg press
2.5–5.0 kg at each side of the bar) for typical 1RM tests is exercise, whereas others found no differences between
lower than those reported using a linear velocity transducer muscle groups (3,19). These discrepancies could be due to
APPLIED SCIENCES

(CV = 0.57%, accuracy = 0.25%, 1000 Hz [32]). These (i) the order of the testing, because 1RM or repetitions to
improvements in the quality of the measure allow us to de- failure tests induce central fatigue that can influence the
tect small but significant effects of caffeine on performance second muscle group tested as suggested by Astorino et al.
(29). Others have reported an increased peak torque after (3), and (ii) the relatively low reliability of the muscle en-
caffeine ingestion using isokinetic or isometric devices. durance tests. We have attempted to reduce the measure-
Astorino et al. (4) found that a 5-mgIkgj1 dose of caffeine ment variability by testing only single explosive actions with
improved isometric power whereas a lower 2 mgIkgj1 had the velocity transducer and with our careful testing protocol.
no effect. Jacobson et al. (20) reported that 7 mgIkgj1 caf- Our results allow us to suggest that caffeine has a larger ergo-
feine ingestion improved peak isometric torque at several genic effect during lower body muscle contractions. Additional
angular velocities, findings that have been recently con- research is needed to identify the underlying mechanism re-
firmed by Bazzucchi et al. (5). Our findings using isoinertial sponsible for these differences between upper and lower body

2190 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
musculature. Warren et al. (37) argued that the muscle activa- sprint ability, it is less likely that it could improve single
tion level during MVC may be lower for the knee extensors muscle actions or very short sprint performance (i.e., 4 s
than for other smaller muscle groups. In these smaller groups, long) where fatigue is not limiting. In support for a local
muscle activation without caffeine is already near 100%, and muscle mechanism, caffeine ingestion has been reported to
thus there is minimal room for caffeine to improve contraction increase the electrically evoked force of small (finger ad-
force. This seems to us a very plausible explanation. ductor [25]), medium (peroneus [35]), and large (quadriceps
Although no subject reached the 2004 urinary caffeine [29]) muscle mass when stimulated at a low frequency
threshold for doping (12 KgImLj1; Fig. 5), some of the (20 Hz). In addition, several studies have failed to show
doses produced side effects that remained 24 h after the trial. increased motor unit activation with caffeine using electro-
One of the findings of the present study was that the pre- stimulation superimposed into a maximal voluntary con-
sence of negative side effects increased markedly with the traction (27,35). In contrast, others have found increased
9-mgIkgj1 caffeine dose (Table 1). To our knowledge, these maximal activation using the twitch interpolation technique
results are novel in the literature for a controlled, double- (22,30). As the motor unit recruitment and firing rate are
blind, crossover design where three incremental doses of likely larger during resistance than during endurance type
caffeine are evaluated. In a descriptive cross-sectional study, activities, it could be hypothesized that caffeine could fur-
Desbrow and Leveritt (11) associated habitual caffeine use ther benefit resistance exercise if a local effect is predomi-
by Ironman athletes with side effects and found very minor nant. However, caffeine ingestion has not been shown to
and infrequent adverse caffeine-related symptoms during increase motor unit firing rates in nonfatigued muscle (22,30).
this long distance endurance event. Their retrospective data Again, a bout of maximal contraction could be limited by
did not allow them to analyze the dose–side effects rela- the capacity to voluntarily activate motor units, which seems
tionship. In the present study, we found gastrointestinal to be improved by caffeine ingestion (22). Thus, it is unclear
problems, headaches, and insomnia appearing with doses at by which mechanism (local or central) caffeine ingestion is
or higher than 6 mgIkgj1 (Table 1). In sport events lasting improving force and power during single muscle actions.
longer than half a day (morning or afternoon), these side In conclusion, in this study, we systematically raised caf-
effects generated by early caffeine ingestion could reduce feine dose while varying the load imposed to large muscle
cognitive and physical performance. The side effects data, groups located in the upper and lower body (BP and SQ). The
together with the muscle strength and mechanical power velocity of movement against those loads was improved
output results, allow us to suggest that only in events where in 83% of the trials with caffeine (4.3%–13.1%) and thus
the sport success depends on force application against high muscle power output. Importantly, as resistance increased
loads, like Olympic weightlifting or the start and the first toward 1RM, a higher dose of caffeine was needed to in-
1–5 cycles or strokes in sprints, would it be advisable to crease MPV (Fig. 2) and power (Fig. 3). Although no sub-
ingest caffeine doses higher than 6 mgIkgj1. ject reached the 2004 urinary caffeine threshold for doping,
Few studies have reported a wide range of load–power the increase in dosage produced side effects like gastroin-
data in these basic resistance training exercises (BP, SQ). testinal problems, anxiety, and headaches that remained 24 h
Figure 3 suggests that the shape of the load–power curve after the trial (Table 1). The practical application for sport
is not affected by caffeine ingestion. Furthermore, our data nutrition and performance is that muscle contractions against
coincide with previous reports (18,33) in showing that heavy loads (75%–90% 1RM) also require a high caffeine
muscle power output is very different in BP (25% 1RM) and dose (9 mgIkgj1) to obtain an ergogenic effect. However,
SQ (75% 1RM) exercises, probably due to the biomechanics explosive, high-velocity low-resistance actions require a much
of the primary muscles recruited in each movement. Finally, lower caffeine dose (3 mgIkgj1), thus avoiding the undesir-
no significant differences were observed in the power output able side effects.
developed against 25%–50% 1RM loads in BP, or against
50%–90% 1RM loads in SQ (Fig. 3). Attending to the load–
The authors thank the collaboration of José Marı́a López Gullón,
power data (Fig. 3), it could be suggested that there is not Ricardo Morán Navarro, Alvaro López Samanes, and Luis Sánchez
one but a range of loads that maximize muscle power output Medina from the High-Performance Sports Center Infanta Cristina, APPLIED SCIENCES
(low loads for BP and moderate loads for SQ), and caffeine University of Murcia, University of Castilla-La Mancha and Research
ingestion does improve power at all loads. In agreement with and Sports Medicine Centre from the Government of Navarre, re-
spectively. They also acknowledge the commitment and dedication
recent reports (33), these results make us wonder whether to the testing of each of the 13 high-performance athletes that par-
perhaps excessive attention has been paid to the question of ticipated in this investigation.
identifying a single load for maximizing power output. No funding was received for this work. The authors report no
conflicts of interest.
While delaying central nerve fatigue could be one of The results of the present study do not constitute endorsement
the mechanisms by which caffeine could improve repeated by the American College of Sports Medicine.

REFERENCES
1. Astorino TA, Martin BJ, Schachtsiek L, Wong K, Ng K. Minimal 2. Astorino TA, Roberson DW. Efficacy of acute caffeine ingestion
effect of acute caffeine ingestion on intense resistance training per- for short-term high-intensity exercise performance: a systematic
formance. J Strength Cond Res. 2011;25(6):1752–8. review. J Strength Cond Res. 2010;24(1):257–65.

CAFFEINE DOSE AND NEUROMUSCULAR RESPONSE Medicine & Science in Sports & Exercised 2191

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
 3. Astorino TA, Rohmann RL, Firth K. Effect of caffeine ingestion 23. Kamimori GH, Karyekar CS, Otterstetter R, et al. The rate of ab-
on one-repetition maximum muscular strength. Eur J Appl Physiol. sorption and relative bioavailability of caffeine administered in
2008;102(2):127–32. chewing gum versus capsules to normal healthy volunteers. Int J
4. Astorino TA, Terzi MN, Roberson DW, Burnett TR. Effect of two Pharm. 2002;234(1–2):159–67.
doses of caffeine on muscular function during isokinetic exercise. 24. Kovacs EMR, Stegen J, Brouns F. Effect of caffeinated drinks on
Med Sci Sports Exerc. 2010;42(12):2205–10. substrate metabolism, caffeine excretion, and performance. J Appl
5. Bazzucchi I, Felici F, Montini M, Figura F, Sacchetti M. Caffeine Physiol. 1998;85(2):709–15.
improves neuromuscular function during maximal dynamic exer- 25. Lopes JM, Aubier M, Jardim J, Aranda JV, Macklem PT. Effect
cise. Muscle Nerve. 2011;43(6):839–44. of caffeine on skeletal-muscle function before and after fatigue.
6. Beck TW, Housh TJ, Schmidt RJ, et al. The acute effects of J Appl Physiol. 1983;54(5):1303–5.
a caffeine-containing supplement on strength, muscular endur- 26. Martin JC, Wagner BM, Coyle EF. Inertial-load method de-
ance, and anaerobic capabilities. J Strength Cond Res. 2006; termines maximal cycling power in a single exercise bout. Med Sci
20(3):506–10. Sports Exerc. 1997;29(11):1505–12.
7. Burke LM. Caffeine and sports performance. Appl Physiol Nutr 27. Meyers BM, Cafarelli E. Caffeine increases time to fatigue by
Metab. 2008;33(6):1319–34. maintaining force and not by altering firing rates during submaxi-
8. Childs E, de Wit H. Subjective, behavioral, and physiological mal isometric contractions. J Appl Physiol. 2005;99(3):1056–63.
effects of acute caffeine in light, nondependent caffeine users.
28. Mohr T, Van Soeren M, Graham TE, Kjaer M. Caffeine ingestion
Psychopharmacology. 2006;185(4):514–23.
and metabolic responses of tetraplegic humans during electrical
9. Cohen J. Statistical Power Analysis for the Behavioral Sciences.
cycling. J Appl Physiol. 1998;85(3):979–85.
2nd ed. New Jersey: Lawrence Erlbaum Associates; 1988. p. 569.
10. Cox GR, Desbrow B, Montgomery PG, et al. Effect of different 29. Mora-Rodriguez R, Garcia Pallares J, Lopez-Samanes A, Fernando
protocols of caffeine intake on metabolism and endurance perfor- Ortega J, Fernandez-Elias VE. Caffeine ingestion reverses the
mance. J Appl Physiol. 2002;93(3):990–9. circadian rhythm effects on neuromuscular performance in highly
11. Desbrow B, Leveritt M. Well-trained endurance athletes’ knowl- resistance-trained men. PLoS One. 2012;7(4):e33807.
edge, insight, and experience of caffeine use. Int J Sport Nutr 30. Plaskett CJ, Cafarelli E. Caffeine increases endurance and attenu-
Exerc Metab. 2007;17(4):328–39. ates force sensation during submaximal isometric contractions.
12. Glaister M, Howatson G, Abraham CS, et al. Caffeine supplemen- J Appl Physiol. 2001;91(4):1535–44.
tation and multiple sprint running performance. Med Sci Sports 31. Reilly T, Edwards B. Altered sleep-wake cycles and physical
Exerc. 2008;40(10):1835–40. performance in athletes. Physiol Behav. 2007;90(2–3):274–84.
13. Glaister M, Patterson SD, Foley P, Pedlar CR, Pattison JR, 32. Sanchez-Medina L, Gonzalez-Badillo JJ. Velocity loss as an in-
McInnes G. Caffeine and sprinting performance: dose responses dicator of neuromuscular fatigue during resistance training. Med
and efficacy. J Strength Cond Res. 2012;26(4):1001–5. Sci Sports Exerc. 2011;43(9):1725–34.
14. Goldstein E, Jacobs PL, Whitehurst M, Penhollow T, Antonio J. 33. Sanchez-Medina L, Perez CE, Gonzalez-Badillo JJ. Importance of
Caffeine enhances upper body strength in resistance-trained women. the propulsive phase in strength assessment. Int J Sports Med.
J Int Soc Sports Nutr. 2010;7:18. 2010;31(2):123–9.
15. Graham TE, Spriet LL. Metabolic, catecholamine, and exercise 34. Seo D-i, Kim E, Fahs CA, et al. Reliability of the one-repetition
performance responses to various doses of caffeine. J Appl Phys- maximum test based on muscle group and gender. J Sports Sci
iol. 1995;78(3):867–74. Med. 2012;11(2):221–5.
16. Green JM, Wickwire PJ, McLester JR, et al. Effects of caf- 35. Tarnopolsky M, Cupido C. Caffeine potentiates low frequency
feine on repetitions to failure and ratings of perceived exertion skeletal muscle force in habitual and nonhabitual caffeine con-
during resistance training. Int J Sports Physiol Perform. 2007; sumers. J Appl Physiol. 2000;89(5):1719–24.
2(3):250–9. 36. Van Thuyne W, Delbeke FT. Distribution of caffeine levels in
17. Greer F, McLean C, Graham TE. Caffeine, performance, and me- urine in different sports in relation to doping control before and
tabolism during repeated Wingate exercise tests. J Appl Physiol. after the removal of caffeine from the WADA doping list. Int J
1998;85(4):1502–8. Sports Med. 2006;27(9):745–50.
18. Izquierdo M, Hakkinen K, Gonzalez-Badillo JJ, et al. Effects of 37. Warren GL, Park ND, Maresca RD, McKibans KI, Millard-Stafford
long-term training specificity on maximal strength and power of ML. Effect of caffeine ingestion on muscular strength and endur-
the upper and lower extremities in athletes from different sports. ance: a meta-analysis. Med Sci Sports Exerc. 2010;42(7):1375–87.
Eur J Appl Physiol. 2002;87(3):264–71.
38. Williams AD, Cribb PJ, Cooke MB, Hayes A. The effect
19. Jacobs I, Pasternak H, Bell DG. Effects of ephedrine, caffeine, and
of ephedra and caffeine on maximal strength and power in
their combination on muscular endurance. Med Sci Sports Exerc.
resistance-trained athletes. J Strength Cond Res. 2008;22(2):
2003;35(6):987–94.
464–70.
20. Jacobson BH, Weber MD, Claypool L, Hunt LE. Effect of caffeine
39. Woolf K, Bidwell WK, Carlson AG. The effect of caffeine as an
APPLIED SCIENCES

on maximal strength and power in elite male athletes. Br J Sports

Med. 1992;26(4):276–80. ergogenic aid in anaerobic exercise. Int J Sport Nutr Exerc Metab.
21. Jenkins NT, Trilk JL, Singhal A, O’Connor PJ, Cureton KJ. 2008;18(4):412–29.
Ergogenic effects of low doses of caffeine on cycling performance. 40. Woolf K, Bidwell WK, Carlson AG. Effect of caffeine as an er-
Int J Sport Nutr Exerc Metab. 2008;18(3):328–42. gogenic aid during anaerobic exercise performance in caffeine
22. Kalmar JM, Cafarelli E. Effects of caffeine on neuromuscular naive collegiate football players. J Strength Cond Res. 2009;
function. J Appl Physiol. 1999;87(2):801–8. 23(5):1363–9.

2192 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright © 2013 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.