Are restaurant menu calorie counts required to be accurate?

Menu labeling rules for large chains require calorie figures with a reasonable basis — the number describes the standard recipe at the standard serving size. The rules don't and can't guarantee the specific plate you're served, because hand-portioned oil, sauce, cheese and sides vary from cook to cook and day to day.

Should I log the menu's calorie number or estimate my actual plate?

For tightly standardized items — a packaged burger, a bottled drink — the menu number is a good log entry. For anything assembled by hand (bowls, curries, salads, pasta), a photo-based estimate of what's actually on the plate captures portion variance the menu number cannot, and you can still sanity-check it against the label.

Why does the same dish differ between locations or on delivery?

Because portioning is human. Different cooks pour different amounts of oil and dressing, delivery kitchens may pack sides and sauces differently to the dine-in spec, and independent restaurants often have no gram-level recipe spec at all. The dish name stays the same; the calories on the plate don't.

Guide

How accurate are restaurant calorie counts?

The number on the menu is a real, regulated figure — for the standard recipe. The plate in front of you was assembled by a person with a ladle. Here's how far the two can drift apart, and the sensible way to log meals out.

What the menu number actually measures

Menu labeling rules for large chains require calorie counts to have a reasonable basis: the figure is derived from the standard recipe at the standard serving size, via lab analysis or nutrition databases. That makes it a genuinely useful number — for the recipe. It is not a measurement of your serving, and the rules don't pretend otherwise. Two truths coexist: the label is honest, and your plate can still differ from it meaningfully.

Where the drift comes from

Free-poured oil. The single biggest wildcard. A cook finishing a pan with an extra glug of olive oil adds real calories that no menu can anticipate — and that you can't see once it's absorbed.
Hand-portioned toppings. Cheese, dressing, sauce, guacamole and granola are scooped, not weighed. A generous hand on any of them moves the total noticeably; on a "healthy" dressed salad, the dressing alone can rival the greens.
Sides and bases. Rice, fries and pasta are ladled by eye. The same dish name can arrive with visibly different amounts of its most calorie-dense component.
Recipe drift. Kitchens tweak recipes, suppliers change, and labels aren't always updated in lockstep.

Chains, independents and delivery

The spectrum runs from standardized to improvised. A packaged chain burger is built from pre-portioned components, so its label tracks the plate closely. A chain bowl or burrito is assembled by hand — same brand, more variance. An independent restaurant often has no gram-level spec at all; any calorie figure is an estimate of a dish that changes with the chef's mood. Delivery adds its own layer: portions packed for transport, sauces on the side or already mixed in, and sometimes a different kitchen altogether operating under the same menu.

How to log restaurant meals sensibly

Standardized item? Use the menu number. It's the best available figure for a spec-built product.
Hand-assembled item? Estimate what's actually on the plate. A photo-based estimate captures the thing the menu can't: this serving's portions. Cal AI: Calorie Scanner reads the plate and returns an ingredient-level breakdown you can edit — if the rice pile is clearly double or you left half the dressing, a one-tap correction beats inheriting the spec's assumptions.
Cross-check when both exist. If your plate estimate and the menu label are close, great. If they're far apart, believe your eyes on portions and the label on hidden ingredients — and log somewhere between.
Tell the log what you can't see. "Probably cooked in butter" as a voice note is more honest than pretending the pan was dry.
Don't agonize. One meal out is a rounding error in a consistent week. The goal is a sane estimate, not an audit.

This is also where the classic database approach shows its limits: apps built on chain databases, like the one we compare in Cal AI: Calorie Scanner vs MyFitnessPal, log the corporate spec faithfully — which is exactly right for standardized items and exactly the problem for hand-assembled ones. If you eat out often, it's worth choosing a tracker with that distinction in mind; our guide to the best AI calorie trackers covers it.

Log the plate, not the spec

Three free AI scans — point it at your next restaurant meal and compare against the menu's number.

App Store Google Play

The verdict

A menu calorie count is the recipe's number, not your plate's number. It's trustworthy for spec-built, standardized items and increasingly approximate as human hands enter the assembly. Log standardized food from the label, estimate hand-assembled food from what's actually served, and remember that portion variance — not label dishonesty — is where most of the gap lives.