Will your Nutrition Facts table pass CFIA's test? How the 80/120 rule works

A food company sends one jar of its product to a lab, the sodium comes back 15% over the label, and someone asks whether the label will fail.

It might. It might not. CFIA does not judge a Nutrition Facts table on one jar, and the way it does judge it is written down in a public document: the Nutrition labelling compliance test. This piece walks through it, because the design of the test decides which lab result answers the question.

Twelve units, three composites

CFIA takes at least twelve consumer units at random from one lot and combines them into three composite sub-samples of at least four units each. The three composites are analysed, and their mean is the estimate of the lot’s nutrient content (compliance test, “Sampling”).

So the number that matters is an average of three measurements, each already an average of four units. One unit, tested once, is a different question.

The 80/120 rule

The main criterion compares that mean with the declared value. For nutrients declared in the Nutrition Facts table:

Nutrient	The mean of three composites must be
Protein, carbohydrate, fibre, and vitamins and minerals that occur naturally (Class II)	at least 80% of the declared value
Calories, fat, saturated fat, trans fat, cholesterol, sodium, sugars, sugar alcohols	not more than 120% of the declared value
Vitamins and minerals added to the food (Class I)	at least 100% of the declared value, with no tolerance

The direction matters. Nutrients people want more of can come in low, down to 80%. Nutrients people are told to limit can come in high, up to 120%. A product with 25% less fat than declared does not fail the fat criterion; a product with 25% more does.

The same tolerances apply to the regulated limits behind nutrient content claims (compliance test, “Tolerances (Criterion 2)”).

Rounding comes first

The declared value is compared after it is adjusted using the Nutrition Facts rounding rules, as a minimum or maximum value those rules allow (Criterion 2). A label that says 140 mg sodium is not tested against exactly 140 mg. If your lab compares a result with the printed number, it is comparing with the wrong number. CFIA’s document has worked examples in its appendices.

Two more rules people miss

Criterion 1, the 50% rule. Each of the three composites, on its own, must be within 50% of the declared value adjusted for rounding: at least half, or at most 1.5 times, depending on the nutrient. A lot with an acceptable mean can still fail on one composite that is far off.

Criterion 3, consistency for added vitamins and minerals. For Class I nutrients, CFIA also checks the spread of the three composites. In the document’s terms, a 10% difference from the mean must fall inside a 99% confidence interval for the standard deviation of the lot. In practice: a fortified product that is mixed unevenly can fail even when the average looks right.

What this means when you send a sample to a lab

1. Send enough units. If you want to know how CFIA would see a lot, send at least twelve units from that lot, and ask the lab to make three composites of four.
2. Ask for the declared value adjusted for rounding to be part of the comparison, not the printed number.
3. Ask how added vitamins and minerals are handled. They have no tolerance, so the method’s precision matters more.
4. Treat a single-unit result as a screen. It can tell you something is wrong. It cannot tell you the lot passes.

Which methods and which lab

CFIA does not require other laboratories to use its methods. It recommends methods from the Official Methods of Analysis of AOAC International where possible, other collaboratively studied methods such as AOCS, AACC or ISO, or in-house methods with adequate validation, and methods validated for the food matrix being analysed (compliance test, Appendix 4). It also notes that labs in Canada are accredited by the Standards Council of Canada or CALA, not by CFIA, and that an accredited lab’s scope lists its methods.

Matrix matters here more than people expect. A fat or fibre method that works on a cereal may not be validated for a sauce or a confection. It is a fair question to ask your lab.

Databases are allowed, and checked differently

CFIA says companies may choose how they develop accurate nutrient data, and a nutrient database is one tool. When CFIA verifies label values, it looks at the company’s controls: record keeping, raw material specifications, company lab analysis, documentation of data sources (compliance test, “Use of data bases”). It also recommends that manufacturers rely on laboratory testing to verify their own declarations (Appendix 4).

A database value for a recipe is a prediction. The compliance test measures what is in the jar.

Sources

• Canadian Food Inspection Agency. Nutrition labelling compliance test: summary, Part 1 (sampling, acceptance criteria, tolerances, use of databases) and Appendix 4 (laboratory issues). Retrieved 2 October 2026.

This is a summary of CFIA’s published test, not a compliance opinion on any product; CFIA enforces the rules. If your label results keep landing close to the limits and you are not sure the testing design is right, tell me the product type on LinkedIn and I will tell you how I would set up the test.