The 20% rule

The FDA's Nutrition Labeling and Education Act (NLEA) allows packaged-food labels to deviate from the actual nutrient content by up to 20% in either direction for most macronutrients and calories. This is not a loophole companies abuse maliciously. It is the legal margin that exists because food is a biological product, manufacturing varies, and measuring every batch to lab precision is impossible at scale.

A bag of pretzels labeled 110 calories per serving might really be 88 to 132. A protein bar that says 20g protein might be 16 to 24. A frozen meal labeled 480 calories could be 384 to 576. The label is a target. The contents are an approximation.

Most reputable brands hit much closer than the 20% margin in practice. Spot tests by independent labs have found that established brands tend to land within ±5-10% of the label on packaged goods with consistent ingredients. The margin grows for foods with high natural variance — produce, meat, anything where the input ingredient itself isn't standardized.

Where the variance comes from

Four main sources, in roughly this order:

Serving size definitions. A label says "1 serving = 30g" but you measure 35g because you eyeballed it. That's a 17% error before you even started. Most label "inaccuracy" is actually serving size error on the user's side.

Ingredient batch variance. A chicken breast in a frozen meal can be 4 oz or 4.6 oz depending on the supplier's cuts that week. Companies build a buffer into the label number to stay within FDA tolerance across the whole production run.

Cooking and preparation. Labels reflect the package contents. If you cook in oil that wasn't in the label calculation, your eaten calories are higher. Same for added butter, sauce, dressing.

Measurement method. Bomb calorimetry (the gold standard for measuring calories in a food) is expensive and not done per batch. Most labels are calculated from ingredient databases plus the formulation recipe, not measured directly. The database itself has variance.

What this means for tracking

If you're a strict packaged-foods tracker — every meal weighed, every label trusted — your daily total is theoretically accurate to within about ±10%. In practice, most people round generously, eyeball portions, and miss small additions, so real-world precision is closer to ±15-20%.

If you're a voice or photo tracker — describing meals to an estimator — your precision is ±15-25% on common foods. The gap between the two is smaller than the marketing for either method suggests.

The practical implication: agonizing over whether your protein bar was "really" 20g or 22g is not the meaningful variable. The meaningful variable is whether you're consistently in the right range over weeks. ±20% on a daily target of 2,200 calories is ±440. That's a lot in absolute terms. Spread over a week of consistent tracking, the trend still shows up clearly because errors are mostly random in both directions.

When the label IS systematically wrong

A few categories where labels have been documented to under-report calories more often than they over-report:

  • Restaurant chain meals with state-mandated nutrition labels (California, NYC) tend to be slightly under-reported on average. Studies have found chain-restaurant calorie counts run about 18% under the actual measured value.
  • "Healthy" branded products — protein bars, granola, low-fat or low-sugar branded items — show wider variance than basic packaged goods. The brand has marketing pressure to keep numbers attractive.
  • Bulk-cooked or prepared deli foods sold by weight at grocery stores often have generic labels that don't reflect the day's actual recipe.

None of this should make you stop trusting labels. It should make you stop expecting precision the labels were never designed to deliver.

What to do about it

Four practical adjustments:

  1. Round generously when in doubt. If a serving size estimate is between two values, take the higher one. That nudges your tracking toward conservative — which catches the systematic under-reporting bias.
  2. Track over weeks, not days. A 200-calorie daily error doesn't matter if you're consistent. A 200-calorie weekly trend does.
  3. Don't sweat the labeled snack. A 110-calorie protein bar is 110 calories for tracking purposes. Even if it's really 130, the difference is noise compared to the meals around it.
  4. Validate against bodyweight, not the math. If you've eaten at a 400-calorie deficit for 4 weeks per the labels and your weight hasn't moved, your effective deficit is smaller than you think — adjust the daily target down 100-200 and recheck. The scale is the ultimate auditor of your tracking accuracy.

The label is a starting point. Your bodyweight trend is the calibration check. If you align those two over time, the label margin stops being a problem and becomes part of the system.