How Accurate Are Nutrition Labels?
How accurate are nutrition labels? Lab tests find calories a few percent over, the FDA's 20% limit runs one way, and restaurant menus follow looser rules.
Counter44
Calorie Tracker
Log your meals as you go
Counter44 tracks calories and macros against a daily target.
Free · No sign-up needed
Disclaimer: This article is general nutrition information, not medical advice. Calorie targets and the numbers in the worked example are estimates. If you have a medical condition or specific dietary needs, or you’re pregnant, talk to a doctor or registered dietitian.
Nutrition labels are usually close, and the legal slack runs one way. A packaged food’s calories can test up to 20% over the label before it counts as misbranded, with no matching number for food that comes in under. In one lab check of 24 snack foods, calories ran a median 4.3% over the label, which the authors called “well within FDA limits.”
- For calories, 20% is only a ceiling. The 80% floor is for naturally occurring protein, most carbs and fiber.
- The FDA tests a blend of 12 units from one lot, so one package can sit further off.
- Menus follow looser rules. In one study, 19% of 269 restaurant items ran at least 100 kcal over.
- Memory is the bigger gap. In a study of 484 adults, 24-hour recalls came in 12–20% under measured energy use.
What the FDA rule actually allows
The rule sits in 21 CFR 101.9(g). The FDA doesn’t test one box. Its sample is “a composite of 12 subsamples (consumer units), taken 1 from each of 12 different randomly chosen shipping cases, to be representative of a lot.”
For calories, the line runs one way. A food is misbranded “if the nutrient content of the composite is greater than 20 percent in excess of the value for that nutrient declared on the label” (g)(5). Coming in under gets no number: “Reasonable deficiencies of calories … under labeled amounts are acceptable within current good manufacturing practice” (g)(6).
So a food labeled 100 calories fails when the composite tests above 120. You’ll often read that the real count could fall anywhere from 80 to 120. For calories, only the 120 is in the rule.
| Declared on the label | What the 12-unit composite must show | Rule |
|---|---|---|
| Calories, total fat, saturated fat, trans fat, cholesterol, sodium, total sugars | No more than 120% of the label. Shortfalls are acceptable within good manufacturing practice. | 101.9(g)(5) and (g)(6) |
| Naturally occurring protein, total carbohydrate, dietary fiber, polyunsaturated and monounsaturated fat, vitamins and minerals | At least 80% of the label | 101.9(g)(4)(ii) |
| Protein, fiber, vitamins and minerals added to the food | At least 100% of the label | 101.9(g)(4)(i) |
Added sugars join the first row when they’re the only sugars in the food. The carb floor holds for most foods, not all. For products such as fruit drinks, juices and confectioneries, with sugars at 90% or more of total carbohydrate, FDA guidance gives carbs the 120% ceiling instead.
The lab gets some slack too
Rule (g)(5) adds that “no regulatory action will be based on a determination of a nutrient value that falls above this level by a factor less than the variability generally recognized for the analytical method used in that food at the level involved.” A result just past 120% that’s within the method’s normal error doesn’t trigger action.
Canada’s compliance test for naturally occurring nutrients is one-sided too, with an 80% floor for nutrients such as protein and fibre and a 120% ceiling for calories, fat, sugars and sodium.
Rounding: why 4 calories can print as 0
Labels don’t print exact amounts. Section 101.9(c) sets the steps:
- Calories: “to the nearest 5-calorie increment up to and including 50 calories, and 10-calorie increment above 50 calories, except that amounts less than 5 calories may be expressed as zero.”
- Fat: to the nearest 0.5 g below 5 g and the nearest gram above. Less than 0.5 g prints as 0.
- Protein and carbohydrate: to the nearest gram. Less than 1 g can read “less than 1 gram,” and less than 0.5 g may print as 0.
Above 50 calories, rounding moves the printed number by 5 kcal at most, up or down. The bottom step works differently: a serving with 4.9 calories may print as 0, so five servings of a “zero-calorie” item can leave nearly 25 calories out of your log.
Why the label’s macros don’t add up to its calories
Multiply a label’s protein and carbs by 4 and its fat by 9, and you’ll often miss the calorie line. Part of the reason is rounding, since the factors “shall be applied to the actual amount (i.e., before rounding)” of each nutrient (c)(1)(i). Take a hypothetical serving with 2.4 g protein, 21.4 g carbs and 7.4 g fat:
- Unrounded: 2.4 × 4 + 21.4 × 4 + 7.4 × 9 = 161.8 kcal, printed as 160.
- From the printed 2 g, 21 g and 7 g: 8 + 84 + 63 = 155 kcal.
Same food, 5 calories apart, and the label is correct.
Labels also don’t have to use plain 4/4/9. Under (c)(1)(i), they can also use specific Atwater factors, take “non-digestible carbohydrates and sugar alcohols” out of total carbs, apply FDA-approved factors for specific foods, or use “bomb calorimetry data subtracting 1.25 calories per gram protein to correct for incomplete digestibility.” The macros to calories calculator runs the 4/4/9 math for any set of grams, so you can see the gap.
How far off labels are in practice
The 20% rule is the legal edge. Lab studies show where labels land, though the newest peer-reviewed US data we found on packaged-food calories is more than a decade old.
Snack foods (Jumpertz, 2013). Researchers measured the “true” caloric content of 24 popular US snack food products. Serving weights ran a median 1.2% above the label, which wasn’t statistically significant. Adjusted for serving size, calories came in a median 4.3% (6.8 kcal) over the label, with the middle half of products between 0.2% and 13.7% over.
The authors’ summary: “Among a convenience sample of energy-dense snack foods, caloric content is higher than stated on the nutrition labels, but overall well within FDA limits.”
Reduced-calorie prepared foods (Urban, 2010). Ten supermarket frozen meals measured an average 8% more energy than stated, and 29 quick-serve and sit-down restaurant foods averaged 18% more. These gaps “did not achieve statistical significance,” because results varied so much between items.
How companies pick the number
FDA guidance leaves manufacturers some choice. Companies “may choose to label the nutrient values on products conservatively, so that the products bearing these labels have a high probability of passing an FDA compliance evaluation.” And “it is the manufacturer’s choice whether to use a mean value or a predicted value on the nutrition label.”
Databases are averages too. USDA FoodData Central calls its values a “snapshot in time,” reflecting “particular foods sampled at particular times from particular locations.” Neither a label nor a database entry measures the food on your plate.
Labels don’t measure what you absorb
This is a separate question from compliance. Atwater factors predict energy from protein, carbs and fat. Feeding studies measure what the body actually gets (metabolizable energy), and for some foods the two disagree.
- Almonds (Novotny, 2012). In 18 healthy adults, “the Atwater factors, when applied to almonds, resulted in a 32% overestimation of their measured energy content.”
- Walnuts (Baer, 2016). In 18 adults, the metabolizable energy of walnuts “was 21% less than that predicted by the Atwater factors.”
Neither result means nut labels break the rules. Label calories come from conversion factors or lab calorimetry, not from measuring what people absorb.
Restaurant menus play by different rules
The 20% test stops at the menu board. On whether the 80/120 rule applies to menu items, the FDA’s restaurant labeling guide answers: “No. It does not apply to standard menu items subject to the requirements of 21 CFR 101.11.” Menu numbers need “a reasonable basis” and must follow the rounding rules. Under 101.11(c)(1), that basis can be “nutrient databases (with or without computer software programs), cookbooks, laboratory analyses, or other reasonable means.”
How do menus hold up? A 2011 JAMA study measured 269 items from 42 restaurants with bomb calorimetry (Urban, 2011). Overall, stated calories were close: a difference of 10 kcal per portion that wasn’t statistically significant. Single items varied more. Fifty of the 269 (19%) measured at least 100 kcal per portion above the stated figure.
The direction depended on the dish. The authors found “understated energy contents for those with lower energy contents,” while higher-calorie foods tended to contain less than stated. Go with the menu number, but if a “light” dish looks generous, round it up.
What label error adds up to across a day
To put the percentages on one scale, take a hypothetical day in round numbers. It doesn’t describe any real product. Say you log 2,000 kcal, and 1,200 of it comes from six packaged servings, each labeled above 50 kcal.
| Source of error on this hypothetical day | Size | What it means |
|---|---|---|
| Rounding on six labels | up to ±30 kcal | ±5 kcal per label at most, in either direction |
| Five “0 calorie” servings | just under 25 kcal at most | each serving under 5 kcal |
| Median gap in the snack study (+4.3%) | about +52 kcal | middle half of products: about +2 to +164 kcal |
| Legal ceiling (+20%) | +240 kcal | the enforcement line for a 12-unit composite, not a typical gap |
| 24-hour recall gap in the OPEN study | about 270–330 kcal (men), 380–500 kcal (women) | a 2,000 kcal log that runs 12–14% (men) or 16–20% (women) under true intake |
Two caveats. The snack study used a convenience sample of energy-dense snacks, so other foods may differ. And rounding errors partly offset, since some labels round up and others down.
The OPEN study compared what 484 adults aged 40 to 69 reported eating with their energy use measured by doubly labeled water. On 24-hour recalls, men underreported by 12–14% and women by 16–20%. Those were recalls, not real-time app logging.
You may also see a 47% underreporting figure. It came from 10 people (nine women and one man) with a history of diet resistance (Lichtman, 1992), not from people in general.
On this hypothetical day, the snack study’s typical label gap comes to a few dozen calories. OPEN’s recall percentages come to a few hundred.
Why consistent logging still works
What follows is our reasoning, not a study finding. Errors that change from day to day, like rounding or a restaurant meal, partly offset over a week. An error that stays the same every day, like a staple food that always runs a bit over its label, doesn’t. It shows up as a gap between what your log says and what your weight does.
So the weight trend is the check. Compare your average logged intake with a few weeks of weight trend, and if they don’t match what your target predicts, adjust the target instead of chasing every label. It was an estimate to begin with.
The research on logging is encouraging, with caveats. A 2011 systematic review reported that “a significant association between self-monitoring and weight loss was consistently found; however, the level of evidence was weak because of methodologic limitations” (Burke). In a 24-week online program with 142 participants, 90.8% of them women, “the frequency of self-monitoring is significantly related to weight loss” (Harvey, 2019). Neither study promises a result for any one person.
Habits that keep the error small and steady
- Weigh in grams. Compliance is “based on the metric measure specified in the label statement of serving size” (g)(7), so copy the label’s gram serving, not the household measure. The food calories by weight calculator scales a label to the grams you ate.
- Keep one method. A steady method gives a steady error, which the weight trend can show.
- Log home cooking by ingredient. How to calorie count homemade food covers weighing each ingredient and the finished pot.
If you track in Counter44, the calorie and macro tracker behind this site, there’s no barcode scanner, so a packaged food goes in as a custom food: enter the label’s serving size in grams or mL, switch to per-serving values and copy its calories and macros. Restaurant meals fit Quick add, which needs only calories. History’s Calories card shows your average over the last 7 days with food logged, and its Weight card’s 30-day trend is free (the 90- and 365-day ranges are part of the paid Pro subscription). You can download Counter44 for iOS and Android.
Labels won’t ever be exact. With a steady method and a weight trend, you can see what they miss and adjust for it.
References
- 21 CFR 101.9, Nutrition labeling of food (Cornell Legal Information Institute). Rounding rules (c)(1), (c)(2), (c)(6) and (c)(7); calorie methods (c)(1)(i); the 12-unit composite (g)(2); the 80% and 100% floors (g)(4); the 20% calorie ceiling (g)(5); acceptable deficiencies (g)(6); and compliance on the metric serving (g)(7).
- U.S. Food and Drug Administration. Guidance for Industry: Guide for Developing and Using Data Bases for Nutrition Labeling. The 120% ratio, conservative labeling, the manufacturer’s choice of a mean or predicted value, and the high-sugar carbohydrate exception.
- U.S. Food and Drug Administration. A Labeling Guide for Restaurants and Retail Establishments Selling Away-From-Home Foods, Part II (PDF). Question 6.21: the 80/120 rule doesn’t apply to standard menu items.
- 21 CFR 101.11, Nutrition labeling of standard menu items (Cornell Legal Information Institute). The reasonable-basis requirement for menu nutrient values (c)(1).
- Jumpertz, R. et al. (2013). Food label accuracy of common snack foods. Obesity, 21(1), 164-169. 24 snack foods: calories a median 4.3% (6.8 kcal) above the label.
- Urban, L. E. et al. (2010). The accuracy of stated energy contents of reduced-energy, commercially prepared foods. Journal of the American Dietetic Association, 110(1), 116-123. Frozen meals averaged 8% and restaurant foods 18% above stated values; not statistically significant.
- Urban, L. E. et al. (2011). Accuracy of stated energy contents of restaurant foods. JAMA, 306(3), 287-293. 269 items from 42 restaurants: accurate overall, with 19% at least 100 kcal per portion over.
- Novotny, J. A. et al. (2012). Discrepancy between the Atwater factor predicted and empirically measured energy values of almonds in human diets. American Journal of Clinical Nutrition, 96(2), 296-301. Atwater factors overestimated almonds’ measured energy by 32%.
- Baer, D. J. et al. (2016). Walnuts consumed by healthy adults provide less available energy than predicted by the Atwater factors. Journal of Nutrition, 146(1), 9-13. Metabolizable energy 21% below the Atwater prediction.
- Subar, A. F. et al. (2003). Using intake biomarkers to evaluate the extent of dietary misreporting in a large sample of adults: the OPEN study. American Journal of Epidemiology, 158(1), 1-13. Underreporting on 24-hour recalls against doubly labeled water.
- Lichtman, S. W. et al. (1992). Discrepancy between self-reported and actual caloric intake and exercise in obese subjects. New England Journal of Medicine, 327(27), 1893-1898. 10 people with a history of diet resistance underreported intake by 47%.
- Burke, L. E., Wang, J. and Sevick, M. A. (2011). Self-monitoring in weight loss: a systematic review of the literature. Journal of the American Dietetic Association, 111(1), 92-102. A consistent association between self-monitoring and weight loss, with weak evidence.
- Harvey, J. et al. (2019). Log often, lose more: electronic dietary self-monitoring for weight loss. Obesity, 27(3), 380-384. 142 participants in a 24-week online program; logging frequency related to weight loss.
- USDA FoodData Central, FAQ page. Database values as a “snapshot in time.”
- Canadian Food Inspection Agency. Nutrition labelling compliance test. Canada’s one-sided 80% and 120% acceptance criteria.
Counter44
Calorie Tracker
Start with one meal
Counter44 works out your daily target from your height, weight and goal. Then you log what you eat and keep a streak going.
Free · No sign-up needed
Frequently Asked Questions
Can a food have more calories than its label says?
Yes, within a limit. Under 21 CFR 101.9(g)(5), a packaged food is misbranded when a blend of 12 units from one lot tests more than 20% above the labeled calories, and the FDA allows for the normal error of the lab method. The rule judges that blend, so a single package can land further from the label.
Is there a limit on a label that lists more calories than the food has?
Not a numeric one. The rule calls "reasonable deficiencies" of calories under the labeled amount acceptable within good manufacturing practice. The 80% floor you may have read about applies to naturally occurring protein, most carbohydrates and fiber, not to calories.
Why does a zero-calorie food still have calories?
US labels may show any serving under 5 calories as 0, and a serving with less than 0.5 g of fat prints as 0 g. One serving barely matters, but they add up: five servings at just under 5 calories each come to nearly 25 calories.
Why don't a label's protein, carbs and fat add up to its calories?
The calorie line is worked out from the unrounded amounts and then rounded, while the grams are rounded separately. Labels may also use specific Atwater factors, subtract non-digestible carbohydrates and sugar alcohols, or use bomb calorimetry data. Running 4/4/9 on the printed grams gives a rough check, not an exact match.
Are restaurant calorie counts as accurate as packaged food labels?
They follow a looser rule. The FDA says the 80/120 rule doesn't apply to standard menu items. Their values need a "reasonable basis" and may be worked out from nutrient databases, cookbooks or lab analysis. In a 2011 study of 269 items from 42 restaurants, stated calories were accurate overall, but 19% of items measured at least 100 kcal per portion more than stated.
How far off are packaged food labels in lab tests?
The average gaps found so far are small. In a study published in 2013, calories in 24 popular US snack foods tested a median 4.3% over the label, which the authors described as well within FDA limits. A 2010 study found 10 reduced-calorie frozen meals averaged 8% over their stated values, a difference that wasn't statistically significant.
Do you absorb all the calories a label lists?
Not always. Label calories come from conversion factors or lab calorimetry, not from measuring what people absorb. In a feeding study of 18 adults, Atwater factors overestimated the measured energy of almonds by 32%. In another of 18 adults, the metabolizable energy of walnuts was 21% less than the factors predicted. Neither finding means those labels broke the rules.
Is calorie counting worth it if labels aren't exact?
In one snack-food study, calories ran a median 4.3% over the label. In a separate study, 24-hour recalls missed 12–20% of measured energy use. Weigh in grams, log the same way every day, and compare your logged average with your weight trend over several weeks. A 2011 review found self-monitoring consistently associated with weight loss, though it rated the evidence weak. Calorie targets are estimates, not medical advice.