How Do You Verify Nutrition Label Compliance Against Lab Analysis?

If your product carries a nutrition facts panel, the numbers on that label are not just marketing claims—they are regulated declarations. In China, GB 28050 sets the rules; in the United States, FDA 21 CFR 101 governs. Both standards require that the labeled values for calories, fat, sodium, carbohydrates, protein, and other nutrients fall within defined tolerance bands when compared to laboratory analysis of the actual product. This article explains what nutrition label compliance means, how to verify it, and where the common pitfalls hide.

What It Is

Nutrition label compliance is the process of comparing the declared nutrient values on a product's label (or nutrition information panel) against the values obtained from laboratory testing of representative samples. The goal is to determine whether the label is "truthful and not misleading" under the applicable regulation.

• China (GB 28050): The standard specifies that the actual nutrient content should not be less than 80% of the labeled value for protein, and not more than 120% of the labeled value for fat, saturated fat, cholesterol, sodium, and sugars. For energy (calories) and carbohydrates, the tolerance is 80%–120% of the labeled value. For vitamins and minerals, the lower limit is 80% of the labeled value (with no upper limit for most, except where a maximum is set).
• USA (FDA 21 CFR 101.9(g)): The FDA uses a "Class I" and "Class II" system. For nutrients added to the product (Class I), the labeled value must be at least the actual value (i.e., you cannot under-declare added vitamins or minerals). For nutrients naturally occurring (Class II), the labeled value must be at least 80% of the actual value for protein, fiber, and certain vitamins/minerals; for calories, fat, saturated fat, cholesterol, sodium, and sugars, the labeled value must not exceed 120% of the actual value (i.e., you cannot over-declare these).

In both systems, the key concept is a tolerance band: the label is compliant if the measured value falls within an acceptable range around the declared value.

How It Works: Formula and Steps

The compliance check is not a single-point comparison. It involves sampling, statistical treatment, and a decision rule.

Step 1 – Sample plan. Take a representative number of samples from different production lots. Both GB 28050 and FDA guidance generally require at least 12 samples from multiple lots for a robust check, though a minimum of 6 is sometimes accepted for preliminary screening.

Step 2 – Laboratory analysis. Analyze each sample for the nutrients of interest using an appropriate official method (e.g., AOAC methods for FDA, GB methods for China). The lab must be accredited and use validated methods.

Step 3 – Calculate the mean and standard deviation of the measured values.

Step 4 – Apply the tolerance rule. The general formula for a one-sided tolerance check (typical for label compliance) is:

For nutrients where the label must not be overstated (e.g., fat, sodium, sugars in both systems):

• Compliant if: Measured mean ≤ Label value × 1.20 (upper tolerance)
• Also check the lower bound if applicable (e.g., for protein, measured mean ≥ Label value × 0.80)

For nutrients where the label must not be understated (e.g., added vitamins in FDA Class I, or protein in China):

• Compliant if: Measured mean ≥ Label value × 0.80 (lower tolerance)

Step 5 – Statistical allowance for sampling error. In practice, regulators allow for natural variation. The FDA uses a "composite" approach: for Class II nutrients, the sample mean must be ≥ 80% of the label value, but they also permit a variability allowance based on the standard deviation. A common conservative rule used by industry is:

• Upper tolerance: Label value × 1.20 ≥ Sample mean + (t × SD / √n)
• Lower tolerance: Label value × 0.80 ≤ Sample mean − (t × SD / √n)

where t is the Student's t-value for n−1 degrees of freedom at 95% confidence (e.g., t ≈ 2.201 for n=12), and SD is the sample standard deviation.

A Worked Illustrative Example

Example data (illustrative only). Suppose you produce a snack bar labeled with Sodium 200 mg per serving (China GB 28050 applies). You analyze 12 samples from 3 lots. The measured sodium values (mg/serving) are:

205, 210, 195, 220, 208, 202, 215, 198, 212, 207, 210, 205

• Mean = 207.3 mg
• SD = 6.8 mg
• n = 12, t(0.95, df=11) = 2.201

Upper tolerance check (sodium must not exceed 120% of label):

• Label upper limit = 200 × 1.20 = 240 mg
• Upper confidence bound = Mean + (t × SD / √n) = 207.3 + (2.201 × 6.8 / √12) = 207.3 + 4.3 = 211.6 mg

Since 211.6 mg < 240 mg, the product passes the upper tolerance.

Lower tolerance check (China GB 28050 also has a lower bound for sodium?)
GB 28050 does not set a lower bound for sodium (only an upper bound of 120%). So no lower check is needed. The label is compliant for sodium.

If this were protein (label 10 g, measured mean 8.5 g, SD 0.5 g, n=12):

• Lower limit = 10 × 0.80 = 8.0 g
• Lower confidence bound = Mean − (t × SD / √n) = 8.5 − (2.201 × 0.5/√12) = 8.5 − 0.32 = 8.18 g

Since 8.18 g > 8.0 g, protein passes. If the mean were 7.9 g, the lower bound would be 7.58 g, which is below 8.0 g—failing compliance.

Common Pitfalls

1. Ignoring the lower bound for nutrients like protein and fiber. Many manufacturers only check the upper limit (e.g., "don't exceed fat"), but GB 28050 and FDA both require minimum content for protein, vitamins, and minerals. Under-declaring protein is a violation.
2. Using a single sample. One lab result is not enough. Variability between lots and within a lot is real; a single sample can easily fall outside the band even when the true mean is compliant.
3. Confusing FDA Class I and Class II. If you add a vitamin, the FDA requires the label to be at least the measured value (no upper tolerance). If you do not add it, you have an 80% lower tolerance. Mixing these up leads to false compliance conclusions.
4. Not using the correct analytical method. Different methods (e.g., for fiber or sugars) can yield different results. Always use the method referenced by the regulation (AOAC for FDA, GB methods for China).
5. Forgetting the "reasonable excess" rule. In the US, for calories and certain nutrients, the FDA allows a small excess above the label (e.g., up to 20% for calories) but not for added vitamins. In China, GB 28050 also has specific rules for "0" claims (e.g., "0 trans fat" requires < 0.3 g per serving). A label that says "0" but measures 0.2 g is non-compliant.

Closing

Nutrition label compliance is a statistical exercise, not a guess. By sampling correctly, applying the correct tolerance band from GB 28050 or 21 CFR 101, and using a confidence bound, you can confidently verify that your labels are defensible. To simplify the math and reduce the risk of manual error, try our free Nutrition Label Compliance Calculator at https://www.6sq.com/tools/nutrilab/—it applies the tolerance rules and statistical checks automatically, so you can focus on making great food, not on spreadsheet formulas.