How Accurate Are Restaurant Menu Calories, Really?

What the research actually found

Independent lab testing of chain restaurant menu items has found stated calorie counts miss actual content by a wide range — commonly cited studies put the gap at roughly 10-50% depending on the dish and the restaurant type. In one widely-cited analysis, about 40% of tested items had at least 10 more calories than the menu claimed, while just over half had at least 10 fewer — meaning menus aren't uniformly optimistic or pessimistic, they're just imprecise in both directions.

The more concerning pattern: items explicitly marketed as "light" or low-calorie were disproportionately likely to be undercounted — exactly the dishes a calorie-conscious diner is most likely to trust at face value.

Why fast food is more consistent than sit-down

Not all restaurants are equally unreliable. Fast food and fast-casual chains — where portions are pre-measured, scooped with standardized tools, and prepared from a fixed recipe — tend to track much closer to their stated numbers. Sit-down restaurants, where a cook is hand-plating and eyeballing portions, show significantly more variance: the same dish can differ meaningfully from one visit, one location, or one cook to the next.

This is a genuinely useful rule of thumb: trust chain fast food numbers a bit more, and build in extra buffer at a sit-down restaurant, especially for dishes with sauces, dressings, or oil that get added "to taste" by hand.

The FDA doesn't lab-test your burrito

No regulator routinely verifies restaurant calorie claims. Under FDA menu-labeling rules, chains with 20+ locations must post calorie counts and have a "reasonable basis" for the number — typically a lab test or a standard nutrient database calculation done once, at launch — but there's no ongoing audit requirement, and no one is re-testing the burrito you actually got handed today. The number on the menu is a good-faith estimate from when the item was designed, not a live measurement of your plate.

What "verified" means for the data behind this site

Every number on AutoCal's chain nutrition pages is checked against the chain's own published nutrition disclosures — the same figures the chain is legally required to have a reasonable basis for — not a crowd-sourced guess or an AI estimate from a photo. That's the accurate baseline available. It carries the same real-world caveats as any restaurant number: your specific plate, on your specific day, from your specific location, can still vary.

How to use estimates without getting burned

Treat any single restaurant number as a range, not a precise fact — add a mental 10-15% buffer, especially at sit-down restaurants or for anything with a sauce spooned on by hand. More importantly, manage your average over a week rather than obsessing over any one meal; small per-meal inaccuracies wash out over time, while consistently choosing lower-calorie, higher-protein items (which is what the healthiest-picks section on every chain page is built to surface) matters far more than chasing decimal-point precision on any single dish.

A quick sanity check you can run yourself

Two tells are worth watching for. First, suspiciously round numbers — an entrée listed at an even 500 or 1,000 calories is more likely to be a rough calculation than a lab result; real measurements tend to land on odd, specific numbers. Second, compare the same protein across formats at the same restaurant — grilled chicken in a bowl versus the same grilled chicken in a sandwich versus a salad. If the protein and calorie numbers scale sensibly with what's obviously different (bun, dressing, rice), the underlying data is probably being tracked carefully. If they look arbitrary or inconsistent, be more skeptical of the fine print.

None of this means distrust the number entirely — it means use it as the strong estimate it is, not a certified lab result, and lean on the pattern across many items rather than any single stated figure.

FAQ