INFORMATION DOCUMENT:PRACTICAL EXAMPLES OF SAMPLING PLANS 1

INFORMATION DOCUMENT ON PRACTICAL EXAMPLES OF SAMPLING PLANS

This Information Document provides help in choosing appropriate sampling plans. These sampling plans are
examples and should not be regarded as prescriptive. Each example is one option for the particular situation.
Commodity committees may find alternative plans that are more appropriate.
Therefore, they do not present fixed values but give reference to correspondent passages of the standards.
The justification of the choice (“why”) of the individual sampling plans and the corresponding decision criteria
ensues from the standards to be used in the individual situations. Usually the determination of the appropriate
sampling plan is unambiguous, a fact, which will help avoid future conflicts between importing and exporting
countries.
The given examples are intended for institutions specializing in sampling and compliance assessment. These
institutions are familiar with the quoted standards (ISO, OIML, ICMSF, etc.) and should be able to understand
the text in spite of the highly condensed presentation.
Sampling and decision concepts include wrong acceptance and wrong rejection of a lot, which are interrelated.
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 2

Examples of Sampling Plans:

The following Table 1 presents the matrix combinations versus measure / provision with the reference codes of the corresponding examples (Table 2).
The third dimension of product form of marketing (packages/bulk material/foodstuff for consumption) is implemented into the particular examples.

Table 1: Code of Examples

Fruits/ fats/oil fish/fishery milk/milk meat/meat natural mineral cereals

vegetables products products products waters

Qualitative/quantitative FV-Q FO-Q F-Q MI-Q M-Q MW-Q C-Q

characteristics/sensory
inspection
food hygiene FV-FH n.r. F-FH MI-FH M-FH MW-FH n.r.

pesticide residues FV-P FO-P n.r. MI-P M-P n.r. C-P

contaminants FV-C1/2 FO-C F-C MI-C M-C MW-C C-C

residues of veterinary n.r. FO-R F-R MI-R M-R n.r. n.r.

drugs
n.r. = not relevant
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 3

Table 2: Example sampling plans

Example Criteria Type of Sampling Plan Sampling and

Decision Reference
Isolated Lots Continuous series of lots

FV-Q Visible defects Attribute Plan Sampling Consumer: Consumer:

in fruits uncertainty not 1. CXG 50 section 3.1, see specifically ISO CXG 50 section 4.2 (table 10) see specifically:
applicable 2859-2:1985 NMKL Procedure No 12, Annex – Section 4
2. (table 5) and Fig.1 (see below) and ISO 2859-
Sampling: [Link]Sampling procedures for inspection by
Procedure A: A plan is identified by the lot attributes — Part 1: Sampling schemes
size, limiting quality (LQ) and the indexed by acceptance
inspection level (unless otherwise quality limit (AQL) for lot-by-lot inspection
specified, level II shall be used). The Sampling:
sampling size (n) is given in table A. Normal inspection: use of a sampling plan with
Procedure B: A plan is identified by the lot an acceptance criterion that has been devised
size, limiting quality (LQ) and the to secure the producer a high probability of
inspection level (unless otherwise acceptance when the process average of the
specified, level II shall be used). The lot is better than the acceptance quality limit.
sampling size (n) is given in table B1 to Normal inspection is used when there is no
B10. reason to suspect that the process average
Decision: differs from an acceptable level. The sample
For given limiting quality (LQ) and number size is taken from Table 1 and Table 2-A.
of samples n, a lot is compliant if the Tightened inspection: use of a sampling plan
number of items with visible defects is less with an acceptance criterion that is tighter than
than the Rejection number Re (Tables A, that for the corresponding plan for normal
D4). inspection. Tightened inspection is invoked
Producer: when the inspection results of a predetermined
ISO 2859-[Link] number of consecutive lots indicate that the
Sampling: process average might be poorer than the
see “Consumer” AQL. The sample size is taken from Table 1
and Table 2-B.
Decision: Reduced inspection: use of a sampling plan
For given LQ corresponding to AQL of with a sample size that is smaller than that for
consumer sampling plan from ISO 2859-1 the corresponding plan for normal inspection
if applicable, Table D5) and number of and with an acceptance criterion that is
samples n, a lot is compliant if the number comparable to that for the corresponding plan
of items with visible defects does not for normal inspection. The discriminatory ability
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 4

exceed the Acceptance number Ac (Table under reduced inspection is less than under
A). normal inspection.
Reduced inspection may be invoked when the
inspection results of a predetermined number
of consecutive lots
Indicate that the process average is better than
the AQL. The sample size is taken from Table
1 and Table 2-C.
Switching rules:
When normal inspection is being carried out,
tightened inspection shall be implemented as
soon as two out of five (or fewer than five)
consecutive lots have been non-acceptable on
original inspection (that is, ignoring
resubmitted lots or batches for this procedure).
When tightened inspection is being carried out,
normal inspection shall be re-instated when
five consecutive lots have been considered
acceptable on original inspection.
The outline of the switching rules is shown in
Figure 1.
Decision:
For given inspection level, Acceptable Quality
Level (AQL) and number of samples n, a lot is
compliant if the number of items with visible
defects is less than not the Rejection number
Re (Tables 1 and 2 e.g. for single sampling ).
Producer:
ISO 2859-[Link] Sampling procedures for
inspection by attributes — Part 1: Sampling
schemes indexed by acceptance
quality limit (AQL) for lot-by-lot inspection
Sampling: see “Consumer”
Decision:
For given inspection level and Acceptable
Quality Level (AQL, a lot is compliant if the
number of items with visible defects does not
exceed the Acceptance number Ac
(e.g. Tables 1 and 2 for single sampling).
NMKL procedure no 12. (Annex - Section 4):

Figure 1: Levels of inspection and the switching between those.

INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 5

Tighten Inspection
No rejections in 5 2 rejections in 5
consecutive lots consecutive lots
Normal Inspection
Start here

No rejection in 10 lots 1 rejection

Reduced
Inspection

MI-Q Fat content in Variables Plan Consumer and Producer:

milk products Prerequisites: ISO 3951-[Link] Sampling procedures for inspection by variables – Part 1: Specification
1. The lots have not for single sampling plans indexed by acceptance quality limit (AQL)
been screened for lot-by-lot inspection for a single quality characteristic and a single AQL
previously for Sampling:
nonconforming items. For the “s” method acceptance sampling plan the sample standard deviation is used, for
2. Continuing series of the “σ” method acceptance sampling plan the presumed value of the process standard
lots of discrete products deviation is used. If there is sufficient evidence from the control charts (e.g. ´autocontrol´)
all that the variability is in statistical control, consideration should be given to switching to the
supplied by one “σ” method. If this appears advantageous, the consistent value of s (the sample standard
producer using one deviation) shall be taken as σ.
production process Normal inspection is used at the start of inspection (unless otherwise designated) and shall
continue to be used during the course of inspection until tightened inspection becomes
3. quality characteristic necessary or reduced inspection is allowed. Tightened inspection shall be instituted when
must be measurable on two lots on original normal inspection are not accepted within any five or fewer successive
a continuous scale lots. Reduced inspection may be instituted after ten successive lots have been accepted
under normal inspection, provided that these lots would have been acceptable if the AQL
4. the measurement had been one step tighter, production is in statistical control.
error is negligible, i.e. In case that switching rules are not applicable, a particular consumer’s risk quality (CRQ)
with a standard associated with a consumer’s risk should be fixed (e.g. Table K1 or K2). In case of very short
deviation  no more series of lots, ISO 2859-2:2010 might be applied, where the fat content of the sample items
than 1/10 of the sample with respect to the limit (taking into account the measurement uncertainty) might be classified
standard deviation s or as attribute (see example FV-Q).
process standard Summary table 1 directs users to the paragraphs and tables concerning any situation with
deviation  which they may be confronted.
In the case that the Sample sizes are given in table A2 for the sample size letters given in Clause 23, Chart A
measurement error is (for agreed and fixed AQL at 95 % probability of acceptance and LQ at 10 % probability of
significant, the sampling acceptance). This should be verified by inspecting the OC curve from among Clause 24,
number n should be Charts B to R relating to this code letter and AQL.
increased by For the “s” method (CXG 50 section 4.3 (table 14) and NMKL Procedure No 12, Annex –
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 6

n*= n(1+2) where section 5 (table 6) see specifically (ISO 3951-1:2013, Clause 15), the procedure for obtaining
 = / ISO 3951- and implementing a plan is as follows.
1:2013, Annex O) a) With the inspection level given (normally this will be II) and with the lot size, obtain the
sample-size code letter using Table A.1.
5. production is stable b) For a single specification limit, enter Table B.1, B.2 or B.3 as appropriate with this code
(under statistical letter and the AQL, and obtain the sample size n and the acceptability constant k. For
control) and the quality combined control of double specification limits when the sample size is 5 or more, find the
characteristic x is appropriate acceptance curve from among Charts s-D to s-R.
distributed according to c) Take a random sample of size n, measure the characteristic x in each item and then
a normal distribution or calculate x, the sample mean and s, the sample standard deviation (see Annex J). Where a
a close approximation contract or standard defines an upper specification limit U, a lower specification limit L, or
to the normal both, the lot can be judged unacceptable without even calculating s if x is outside the
distribution specification limit(s).
For the “σ” method (CAC GL 50 section 4.3 (table 17) and NMKL Procedure No 12, Annex
– section 5 (table 7)), see specifically (ISO 3951-1:2013, Clause 16) the procedure for
obtaining and implementing a plan is as follows.
a) From Table A.1 the sample-size code letter is obtained.
b) Depending on the severity of inspection, enter Table C.1, C.2 or C.3 with the sample-size
code letter and the specified AQL to obtain the sample size n and acceptability constant k.
c) Take a random sample of this size, measure the characteristic under inspection for all
items of the sample and calculate the mean value.
The sample standard deviation s should also be calculated, but only for the purpose of
checking the continued stability of the process standard deviation (see ISO 3951-1:2013,
Clause 19).
Decision:
a lot is compliant if the average fat content of sample items does not fall below the minimum
value fixed by AQL and LQ taking into account the corresponding standard deviation (s or )
and acceptability constant K. The acceptability constant is given in tables B1 to B3 (s-
method) and C1 to C3 (-method).
If single upper or lower specification limits (U or L) are given, calculate the quality statistic
QU=(U-x)/s or QL=(x-L)/s
where x the sample mean and s, the sample standard deviation.
The lot is acceptable if
QU k or QL k respectively.
For the “σ” method, s must be replaced by σ
FO-Q water content in Variables Plan Consumer and Producer:
butter Prerequisites: see see MI-Q
example MI-Q Sampling:
see example MI-Q
Decision:
a)Microorganisms in Foods 2. Sampling for microbiological analysis: Principles and specific applications. 1986. 2nd Ed. International Commission on Microbiological
Specifications for Foods.
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 7

A lot is compliant if the average water content of sample items does not exceed the maximum
value fixed by AQL taking into account the corresponding standard deviation (s or ) and
acceptability constant k.
See also example MI-Q
F-Q Net weight in Special Plan Consumer and Producer:
prepackaged OIML R 87 (Edition 2004)b): Quantity of product in prepackages
fish Sampling:
see Table 1: Sampling plans for prepackages
Decision:
for fixed ‘Risk Type’ (according to fixed AQL given in OIML R 87) the lot is accepted if all of
the following criteria are met:
1. The average actual quantity of product in a package is at least equal to the nominal
quantity, which is evaluated in the following way:
The total error of the quantity of product in a package is given by the sum of the differences
between the individual product weights and the nominal weight. The average error is given
by that total error divided by the sample size.
The lot is accepted if the average error is a positive number. In case of a negative number,
the lot is accepted if the standard deviation of the individual product weights times the sample
correction factor of Table 1 is higher than the absolute value of the average error.
2. The number of packages containing an actual quantity less than the nominal quantity
minus the tolerable deficiency (Table 2) is less or equal the Number of packages in a sample
allowed to exceed the tolerable deficiencies (Table 1).
3. No package contains an actual quantity less than the nominal quantity minus twice the
tolerable deficiency.
M-Q Nonmeat protein Variables Plan Consumer and Producer:
in meat products Prerequisites: see see MI-Q
example MI-Q Sampling:
see example MI-Q
Decision:
A lot is compliant if the average content of nonmeat protein of sample items does not exceed
the maximum value fixed by AQL taking into account the corresponding standard deviation
(s or ) and acceptability constant k.
See also example MI-Q
MW-Q Sodium content Variables Plan Consumer and Producer:
of prepackaged Prerequisites: see see MI-Q
mineral water example MI-Q Sampling:
see example MI-Q
Decision:
A lot is compliant if the average sodium content of sample items does not exceed the
maximum value fixed by AQL taking into account the corresponding standard deviation (s or
) and acceptability constant k.
See also example MI-Q
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 8

C-Q Moisture in rice Variables Plan on Bulk Consumer and Producer:

grains Material CXG 50 section 5, see specifically: ISO 1[Link] Acceptance sampling plans and
Sampling uncertainty procedures for the inspection of bulk materials / ISO 11648-[Link] Statistical aspects of
implemented sampling from bulk materials — Part 1: General principles / ISO 24333:2009 Cereals and
cereal products -- Sampling
Sampling:
see example C-C
Decision:
for a given maximum limit, the lot is accepted if the sample grand average of these results x
‾
is lower than an upper acceptance value Ux ‾ = mL +  D
FV-FH E. coli in frozen Three-class attributes CXG 50 section 3.2 and NMKL procedure no 12 Annex sampling plans, Section 3, Table 3
vegetables and Plan and Table 4. See specifically: ICMSF (1986) a): Chapter 18 Sampling plans for vegetables,
fruits fruits, and nuts
Sampling:
See Table 28: Sampling plans and recommended microbiological limits for vegetables,
fruits, nuts, and yeast
Decision:
The lot is accepted if not more than 2 items of 5 samples show the presence of E. coli with
a concentration between 100 and 1000 CFU/g. The lot is rejected in the opposite case.
M-FH Staphylococcus Three-class attributes Consumer and Producer:
aureus in fresh Plan CXG 50 section 3.2 and NMKL Procedure No 12, Annex – section 3 (tables 1 and 2), see
or frozen poultry specifically: ICMSF (1986)a): Chapter 13 Sampling Plans For Poultry And Poultry Products
meat Sampling:
see Table 22: Sampling plans and recommended microbiological limits for poultry and poultry
products
Decision:
The lot is accepted if not more than 1 item of 5 samples shows the presence of
Staphylococcus aureus with a concentration between 1000 and 10.000 CFU/g. The lot is
rejected in the opposite case.
F-FH Listeria Two-class attributes Consumer and Producer:
monocytogenes Plan CXG 50-2004 section 3.2 and NMKL Procedure No 12, Annex – section 3 (tables 3 and 4),
in smoked fish – see specifically CXS 311-2013 Standard for smoked fish, smoke-flavoured fish and smoke-
ready-to-eat dried fish, section 6.4.
Sampling: See CXG 61-2007 Guidelines on the application of general principles of food
hygiene to the control of listeria monocytogenes in foods - Annex II Table 1 and 2

Decision: See CXG 61-2007 Guidelines on the application of general principles of food
hygiene to the control of listeria monocytogenes in foods - Annex III
MI-FH Staph. aureus in Two-class attributes Consumer and Producer:
cheese, ‘hard’ Plan CXG 50 section 3.2, see specifically: ICMSF (1986)a): Chapter 15 Sampling plans for milk
and ‘semi-soft’ and milk products
types
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 9

Sampling:
see Table 24: Sampling plans and recommended microbiological limits for dried milk and
cheese
Decision:
The lot is accepted if no item out of 5 samples show the presence of Staph. aureus in 1g,
where the concentration is higher than 10.000 CFU/g. The lot is rejected in the opposite
case.
MW-FH Microorganisms Two-class attributes Consumer and Producer:
in natural Plan CXC 33-1985: Code of Hygienic Practice for Collecting, Processing and Marketing of Natural
mineral water Mineral Waters
(see also ICMSF (1986)a): Chapter 25: Sampling plans for natural mineral waters, other
bottled waters, process waters, and ice.)
Sampling and Decision:
Annex I: Microbiological Criteria, Table: Microbiological Criteria, Point of application: at
source, during production and end product. Assuming a log normal distribution and an
analytical standard deviation of 0.25 log cfu/ml, the sampling plans would provide 95%
confidence that a lot of water containing a defined not acceptable geometric mean
concentration of specific microorganisms would be detected and rejected based on any of
five samples testing positive.
FV-P Pesticides Variables Plan Consumer and Producer:
residues in sampling uncertainty CXG 33-1999: Recommended Methods Of Sampling For The Determination Of Pesticide
apples for not applicable Residues For Compliance With MRLS
compliance with Sampling:
MRL The minimum number of primary samples to be taken from a lot is determined from Table
1b. The primary samples must contribute sufficient material to enable all laboratory samples
to be withdrawn from the bulk sample. The position from which a primary sample is taken in
the lot should preferably be chosen randomly but, where this is physically impractical, it
should be from a random position in the accessible parts of the lot.
The primary samples should be combined and mixed well, if practicable, to form the bulk
sample. The minimum size of each laboratory sample is given by Table 4, 1.2. The analytical
sample should be comminuted, if appropriate, and mixed well, to enable representative
analytical portions to be withdrawn. The size of the analytical portion should be determined
by the analytical method and the efficiency of mixing.
Decision:
The lot complies with a MRL (Pesticide Residues in Food and Feed, Codex Pesticides
Residues in Food Online Database, FAO and WHO 2013) where the MRL is not exceeded
by the analytical result(s). Where results for the bulk sample exceed the MRL, a decision
that the lot is non-compliant must take into account: (i) the results obtained from one or more
laboratory samples, as applicable; and (ii) the accuracy and precision of analysis, as
indicated by the supporting quality control data.
b)International Organization of Legal Metrology (OIML), Bureau International de Métrologie Légale 11, rue Turgot - 75009 Paris - France, Publication OIML R 87 Edition
2004 (E)
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 10

FO-P Pesticides Variables Plan Consumer and Producer:

residues in sampling uncertainty CXG 33-1999: Recommended Methods Of Sampling For The Determination Of Pesticide
vegetable oils not applicable Residues For Compliance With MRLS
Sampling:
The minimum number of primary samples to be taken from a lot is determined from Table
1b. The primary samples must contribute sufficient material to enable all laboratory samples
to be withdrawn from the bulk sample. The position from which a primary sample is taken in
the lot should preferably be chosen randomly but, where this is physically impractical, it
should be from a random position in the accessible parts of the lot.
The primary samples should be packaged units, or units taken with a sampling device. They
should be combined and mixed well, if practicable, to form the bulk sample. The minimum
size of each laboratory sample (0.5 l or 0.5 kg) is given by Table 4, 5.4. The analytical sample
should be comminuted, if appropriate, and mixed well, to enable representative analytical
portions to be withdrawn. The size of the analytical portion should be determined by the
analytical method and the efficiency of mixing.
Decision:
see FV-P
MI-P Pesticides Variables Plan Consumer and Producer:
residues in sampling uncertainty CXG 33-1999: Recommended Methods Of Sampling For The Determination Of Pesticide
cheeses, not applicable Residues For Compliance With MRLS
including Sampling:
processed The minimum number of primary samples to be taken from a lot is determined from Table
cheeses 1b. The primary samples must contribute sufficient material to enable all laboratory samples
units 0.3 kg or to be withdrawn from the bulk sample. The position from which a primary sample is taken in
greater the lot should preferably be chosen randomly but, where this is physically impractical, it
should be from a random position in the accessible parts of the lot.
Whole unit(s) or unit(s) of the primary samples should be cut with a sampling device.
Cheeses with a circular base should be sampled by making two cuts radiating from the
centre. Cheeses with a rectangular base should be sampled by making two cuts parallel to
the sides. The minimum size of each laboratory sample (0.5 kg) is given by Table 5, 3.3. The
analytical sample should be comminuted, if appropriate, and mixed well, to enable
representative analytical portions to be withdrawn. The size of the analytical portion should
be determined by the analytical method and the efficiency of mixing.
Decision:
see FV-P
M-P Fat soluble Variables Plan Consumer and Producer:
pesticides Sampling uncertainty CXG 33-1999: Recommended Methods Of Sampling For The Determination Of Pesticide
residues in not applicable Residues For Compliance With MRLS
cattle carcass Sampling:
for compliance The minimum number of primary samples to be taken from a lot is determined from Table
with MRL 1a, or Table 2 (in the case of a suspect lot). The position from which a primary sample is
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 11

taken in the lot should preferably be chosen randomly but, where this is physically
impractical, it should be from a random position in the accessible parts of the lot.
Each primary sample is considered to be a separate bulk sample. The Minimum size of each
laboratory sample is given in Table 3, 2.1. The analytical sample should be comminuted, if
appropriate, and mixed well, to enable representative analytical portions to be withdrawn.
The size of the analytical portion should be determined by the analytical method and the
efficiency of mixing.
Decision:
see FV-P
C-P Pesticides Consumer and Producer:
residues in rice CXG33-1999: Recommended Methods Of Sampling For The Determination Of Pesticide
grains Residues For Compliance With MRLS
Sampling:
The minimum number of primary samples to be taken from a lot is determined from Table
1b. The primary samples must contribute sufficient material to enable all laboratory samples
to be withdrawn from the bulk sample. The position from which a primary sample is taken in
the lot should preferably be chosen randomly but, where this is physically impractical, it
should be from a random position in the accessible parts of the lot. Sampling devices
required for grain are described in ISO recommendations.
The primary samples should be combined and mixed well, if practicable, to form the bulk
sample. The minimum size of each laboratory sample (1 kg) is given by Table 4, 2. The
analytical sample should be comminuted, if appropriate, and mixed well, to enable
representative analytical portions to be withdrawn. The size of the analytical portion should
be determined by the analytical method and the efficiency of mixing.
Decision:
see FV-P
FV-C1 Aflatoxin in Variables Plan on Bulk Consumer and Producer:
ready-to-eat Material CXS 193-1995: General Standard For Contaminants And Toxins In Food And Feed
treenuts Sampling, sample Sampling:
preparation, and See ANNEX 2. Each lot, which is to be examined for aflatoxin, must be sampled separately.
analytical variances Lots larger than 25 tonnes should be subdivided into sublots to be sampled separately. If a
used to compute lot is greater than 25 tonnes, the number of sublots is equal to the lot weight in tonnes divided
operating characteristic by 25 tonnes. It is recommended that a lot or a sublot should not exceed 25 tonnes. The
curves minimum lot weight should be 500 kg. Representative sampling should be carried out from
the same lot.
In the case of static lots of treenuts contained either in a large single container or in many
small containers, it is not ensured that the contaminated treenut kernels are uniformly
dispersed throughout the lot. Therefore, it is essential that the aggregate sample be the
accumulation of many small incremental samples of product selected from different locations
throughout the lot. The minimum number of incremental samples, the minimum incremental
sample size and the minimum aggregate sample size depend on the lot weight and are given
by Table 1.
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 12

In the case of dynamic lots, the samples are taken from a moving stream of treenuts. The
size of the aggregate sample depends on the lot size, the flow rate of the moving stream
and the parameters of the sampling device.
Two laboratory samples each of 10kg are taken from the aggregate sample. The laboratory
samples should be finely ground and mixed thoroughly. The test portions taken from the
comminuted laboratory samples by a random process should be approximately 50 grams.
Decision:
If the aflatoxin test result is less than or equal to 10 μg/kg total aflatoxin in the test samples
from both laboratory samples, the lot is accepted.
FV-C2 Total aflatoxins Variables Plan on Bulk Consumer and Producer:
in peanuts Material CXS 193-1995: General Standard For Contaminants And Toxins In Food And Feed
intended for Sampling, sample Sampling:
further preparation, and See Aflatoxins Total, Annex 1: Each lot which is to be examined must be sampled separately.
processing analytical variances Large lots should be subdivided into sublots to be sampled separately. The weight or number
used to compute of sublots depend on the lot size and is laid down in Table 1. The number of incremental
operating characteristic samples to be taken depends also on the weight of the lot, with a minimum of 10 and a
curves maximum of 100 (Table 2).
For the sampling procedure see example FV-C1.
The weight of the incremental samples should be approximately 200 grams or greater,
depending on the total number of increments, to obtain an aggregate sample of 20 kg. The
laboratory sample may be a portion of or the entire aggregate sample. If the aggregate
sample is larger than 20 kg, a 20 kg laboratory sample should be removed in a random
manner from the aggregate sample. A minimum test portion size of 100 g should be taken
from the finely ground and mixed laboratory sample.
Decision:
If the aflatoxin test result is less than or equal to 15 μg/kg total aflatoxin in the test sample,
the lot is accepted.
FO-C Erucic acid in Consumer and Producer:
vegetable Oil CXG 50 section 5, see specifically: ISO 1[Link] Acceptance sampling plans and
(bulk) procedures for the inspection of bulk materials / ISO 11648-[Link] Statistical aspects of
sampling from bulk materials — Part 1: General principles
Sampling:
see example C-C
Decision:
see example C-C
for a given maximum limit mL, the lot is accepted if the sample grand average of these results x ‾
is lower than an upper acceptance valuex ‾ = mL +  D.
x
U
F-C Dioxins and Variables Plan Consumer and Producer:
dioxin like Sampling uncertainty ISO 3951-[Link] Sampling procedures for inspection by variables – Part 1: Specification
PCB´s in Fish implemented for single sampling plans indexed by acceptance quality limit (AQL)
(individual for lot-by-lot inspection for a single quality characteristic and a single AQL
Sampling:
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 13

packages or Since the Dioxin content usually is not process controlled, for the “s” method (CXG 50 section
units) 4.3 (table 14) and NMKL Procedure No 12, Annex – section 5 (table 6)) see specifically (ISO
3951-1:2013, Clause 15), the procedure for obtaining and implementing a plan is as follows.
a) With the inspection level given (normally this will be II) and with the lot size, obtain the
sample-size code letter using Table A.1.
b) For a single specification limit U (the ML for Dioxins and dioxin like PCB's), enter Table
B.1, B.2 or B.3 as appropriate with this code letter and the (usually low) AQL, and obtain the
sample size n and the acceptability constant k.
c) Take a random sample of size n, measure the characteristic x in each item and then
calculate 𝑥̅ , the sample mean and s, the sample standard deviation (see Annex J).
Decision:
calculate the quality statistic
QU=(U-𝑥̅ )/s
The lot is acceptable if
Q U k
MI-C Aflatoxin M1 in Consumer and Producer:
Milk (bulk) CXG 50 section 5, see specifically: ISO 1[Link] Acceptance sampling plans and
procedures for the inspection of bulk materials / ISO 11648-[Link] Statistical aspects of
sampling from bulk materials — Part 1: General principles
CXS 193-1995: General Standard For Contaminants And Toxins In Food And Feed
Sampling:
see example C-C
Decision:
see example C-C
for the given maximum limit m L=0.5 g/kg (CXS 193-1995: General Standard for
Contaminants and Toxins in Food and Feed), the lot is accepted if the sample grand average
of these results x ‾ is lower than an upper acceptance value Ux ‾ = mL +  D.
M-C benzo(a)pyrene Variables Plan Consumer and Producer:
in meat Sampling uncertainty ISO 3951-[Link] Sampling procedures for inspection by variables – Part 1: Specification
implemented for single sampling plans indexed by acceptance quality limit (AQL)
for lot-by-lot inspection for a single quality characteristic and a single AQL
Sampling:
see Mi-Q
Sample sizes are given in table A2 for the sample size letters given in Clause 23, Chart A
(for agreed and fixed AQL at 95 % probability of acceptance and LQ at 10 % probability of
acceptance). This should be verified by inspecting the OC curve from among Clause 24,
Charts B to R relating to this code letter and AQL.
3. For the “s” method (CXG 50 section 4.3 (table 14) and NMKL Procedure No 12, Annex –
section 5 (table 6)) see specifically (ISO 3951-1:2013, Clause 15),
The procedure for obtaining and implementing a plan is as follows.
a) With the inspection level given (normally this will be II) and with the lot size, obtain the
sample-size code letter using Table A.1.
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 14

b) Enter Table B.1, B.2 or B.3 as appropriate with this code letter and the AQL, and obtain
the sample size n and the acceptability constant k.
c) Take a random sample of size n, measure the characteristic x in each item and then
calculate 𝑥̅ , the sample mean and s, the sample standard deviation (see Annex J). Where a
contract or standard defines an upper specification limit U, the lot can be judged
unacceptable without even calculating s if 𝑥̅ exceeds the specification limit.
For the “σ” method (CAC GL 50 section 4.3 (table 17) and NMKL Procedure No 12, Annex
– section 5 (table 7)), see specifically (ISO 3951-1:2013, Clause 16) the procedure for
obtaining and implementing a plan is as follows.
4.
5. a) From Table A.1 the sample-size code letter is obtained.
6.
7. b) Depending on the severity of inspection, enter Table C.1, C.2 or C.3 with the sample-
size code letter and the specified AQL to obtain the sample size n and acceptability constant
k.
8.
c) Take a random sample of this size, measure the characteristic under inspection for all
items of the sample and calculate the mean value.
The sample standard deviation s should also be calculated, but only for the purpose of
checking the continued stability of the process standard deviation (see ISO 3951-1:2013,
Clause 19).
Decision:
calculate the quality statistic
QU=(U-𝑥̅ )/s
The lot is acceptable if
Q U k
For the “σ” method, s must be replaced by σ
MW-C Arsenic in Variables Plan on Bulk Consumer and Producer:
Natural Mineral Material CXG 50 section 5, see specifically: ISO 1[Link] Acceptance sampling plans and
Water Sampling uncertainty procedures for the inspection of bulk materials / ISO 11648-[Link] Statistical aspects of
implemented sampling from bulk materials — Part 1: General principles
CXS 193-1995: General Standard For Contaminants And Toxins In Food And Feed
Sampling:
see example C-C
Decision:
see example C-C
for the given maximum limit m L=0.01 mg/kg (CXS 193-1995: General Standard for
Contaminants and Toxins in Food and Feed), the lot is accepted if the sample grand
average of these results x ‾ is lower than an upper acceptance value Ux‾ = mL +  D.
C-C Cadmium Variables Plan on Bulk Consumer and Producer:
content in wheat Material CXG 50 section 5, see specifically: ISO 1[Link] Acceptance sampling plans and
procedures for the inspection of bulk materials / ISO 11648-[Link] Statistical aspects of
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 15

Sampling uncertainty sampling from bulk materials — Part 1: General principles/ ISO 24333:2009 Cereals and
implemented cereal products -- Sampling
Sampling:
sampling from a commodity is classified into two different procedural types:
 sampling of bulk materials for the accurate estimation of an average value of the
quality characteristic assessed in the lot by suppliers
 inspection procedure for bulk materials for making a decision concerning lot
acceptance by consumers.
ISO 11648 is an International Standard for the first type of procedure, ISO 10725 for the
second type, which is based on the assumption that the value of the individual standard
deviation of the specified quality characteristic is known and stable.
The sample size can be estimated using Tables 3 - 22 of the standard ISO 10725:2000 with
fixed producer’s risk  and consumer’s risk  and fixed cost ratio level from the relative
standard deviations dI = I/D and dT = T/D (ISO 10725:2000, 6.3.4) with the sampling
increment standard deviation I and test sample standard deviation T. The number 2nI
increment samples should be taken from the lot and each two of them should be pooled to
two composite samples. From each of the two composite samples 2n T test samples should
be prepared (e.g. homogenized).
For imprecise standard deviations, one measurement per test sample should be performed
(ISO 10725:2000, [Link]).
As an alternative, the number and size of the increment samples and of the test samples are
given in ISO 24333 Table 1 or Table 2 for flowing or static bulk material respectively. That
standard also gives information on suitable sampling devices.
Decision:
As emphasized above, prerequisite is the determination of the estimation standard deviation
E (ISO 10725:2000, 6.2.7 / ISO 11648-1:2003) by monitoring of the cadmium content and
to assess that it is stable. It is permitted to use the values of standard deviations specified
by an agreement between the supplier and the purchaser (e.g. ´autocontrol´) (ISO
10725:2000, 6.2.1).
Taking into account the discrimination interval D = (K K)E (formula C6 in C.4.2) and
assuming that the measurement standard deviation is negligible compared to E (which
should be proven), the following four quantities might be fixed by agreement: the acceptance
quality limit for the lot mean m A (corresponding to AQL, producers’ risk), the probability  of
wrongly rejecting a conforming lot, the non-acceptance quality limit for the lot mean m R
(corresponding to LQ, consumers’ risk), and the probability  of wrongly accepting a
nonconforming lot.
For a given acceptance quality limit m A, the lot is accepted if the sample grand average of
these results x ‾ is lower than an upper acceptance value Ux‾ = mA +  D with the constant for
obtaining the acceptance value  = K/ (K + K).
FO-R Residues of Variables Plan sampling Consumer and Producer:
veterinary drugs uncertainty not CXG 71-2009: Guidelines For The Design And Implementation Of National Regulatory Food
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 16

in fat applicable Safety Assurance Programme Associated With The Use Of Veterinary Drugs In Food
Producing Animals
Sampling: See example F-R, The minimum quantity required for laboratory samples is 500
g (Table A II Group 031).
Decision: see example F-R
F-R Residues of Variables Plan Consumer and Producer:
veterinary drugs Sampling uncertainty CXG 71-2009: Guidelines For The Design And Implementation Of National Regulatory Food
in packaged fish not applicable Safety Assurance Programme Associated With The Use Of Veterinary Drugs In Food
Producing Animals
Sampling:
For non-suspect lots a statistically-based, unbiased sampling program is recommended
(sampling is conducted at random throughout the lot under inspection, although often
systematic sampling is employed). In stratified random sampling the consignment is divided
into non-overlapping groups or strata e.g. geographical origin, time. A sample is taken from
each stratum. In systematic sampling units are selected from the population at a regular
interval (e.g., once an hour, every other lot, etc.). Where non-compliant results are detected
it is possible to derive a crude estimate of the likely prevalence in the general product
population (e.g. ´autocontrol´). The number of primary samples required to give a required
statistical assurance can be read from Appendix A, Table 4.
For exact or alternative probabilities to detect a non-compliant residue, or for a different
incidence of non-compliance, the number of samples n to be taken may be calculated from:
n = ln(1-p) / ln(1-i)
Where p is the probability to detect a non-compliant residue (e.g. 0.95), it is the supposed
incidence of non-compliant residues (e.g. 0.10) in the lot.
In biased or estimated worst case sampling, investigators use their judgment and experience
regarding the population, lot, or sampling frame to decide which primary samples to select.
Such directed or targeted sampling protocols on a sub-population (biased sampling) are
designed to place a greater intensity of inspection/audit on suppliers or product considered
to possibly have a greater potential than the general population of being non-compliant. If
compliant results from biased sampling confirm non-biased program results, they provide
increased assurance that the system is working effectively.
The canned or packaged product should not be opened for sampling unless the unit size is
at least twice the amount required for the final laboratory sample. The final laboratory sample
should contain a representative portion of juices surrounding the product. The minimum
quantity required for laboratory samples is 500 g of edible tissue (Table C VII Class B – Type
08, A).
Decision:
For purposes of control, the maximum residue limit for veterinary drugs (MRLVD) is applied
to the residue concentration found in each laboratory sample taken from a lot. Lot compliance
with a MRLVD is achieved when the mean result for analysis of the laboratory test portions
does not indicate the presence of a residue, which exceeds the MRLVD. Regulatory action
is only taken on samples containing residues, which can be demonstrated to exceed the
INFORMATION DOCUMENT: PRACTICAL EXAMPLES OF SAMPLING PLANS 17

regulatory action limit with a defined statistical confidence.

Mi-R Residues of Variables Plan on Bulk Consumer and Producer:
veterinary drugs Material CXG 71-2009: Guidelines For The Design And Implementation Of National Regulatory Food
in raw milk Sampling uncertainty Safety Assurance Programme Associated With The Use Of Veterinary Drugs In Food
not applicable Producing Animals
Sampling:
See example F-R, The minimum quantity required for laboratory samples is 500 mL (Table B
I Group 033).
Decision:
See example F-R
M-R Residues of Variables Plan Consumer and Producer:
veterinary drugs sampling uncertainty CXG 71-2009: Guidelines For The Design And Implementation Of National Regulatory Food
in meat/meat not applicable
Safety Assurance Programme Associated With The Use Of Veterinary Drugs In Food
products
Producing Animals
Sampling: See example F-R, The minimum quantity required for laboratory samples is 500 g
(Table A I Group 030).
Decision: See example F-R