How to Track Calories When You're Eating Out: Straight Answers to the Nine Questions Everyone Asks

Almost everyone who tracks food is fine on a Tuesday at home and falls apart on a Friday at a restaurant. The barcode scanner is useless, the menu has no numbers, you don't know what the kitchen did with the butter, and by the time dessert arrives you've decided to just skip logging today. That skipped day is the actual problem — not the meal.

So here are direct answers to the questions people genuinely search for when the plate is already in front of them.

Can I trust the calorie numbers printed on the menu?

Directionally yes, precisely no — and the errors are not random. The landmark study, published in JAMA by researchers at the Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts, laboratory-tested 269 food items from national chain restaurants against their stated calories. On average the stated numbers were roughly 18% lower than measured, and 19% of individual items were understated by more than 100 calories.

The most useful finding was where the error concentrated: the lowest-calorie items on the menu were the worst offenders, averaging around 100 calories more than advertised. The diet options are the least trustworthy numbers on the board — which makes sense, because a small absolute slip of oil is a large relative error on a 350-calorie salad.

A 2024 assessment of England's out-of-home calorie-labelling policy found the same broad pattern persists under a national mandate: labelling improves the situation a lot compared with no information, without making individual numbers exact.

There are no numbers at all. What do I log?

Build the plate from components instead of hunting for the dish. Almost every restaurant meal decomposes into four things you can estimate: a protein, a starch, vegetables, and added fat. Estimate each by hand-size — palm of protein, cupped hand of starch, fist of vegetables — and then handle the fat separately, because fat is where restaurant meals diverge from home cooking.

The widely used heuristic is to add one tablespoon of oil or butter — about 120 calories — to any dish that wasn't steamed, poached, grilled dry or served raw. Add two for anything described as pan-fried, crispy, sautéed, glazed, creamy, or served in its own sauce. Restaurant kitchens use fat for flavour and finish in quantities that would look alarming in a home pan, and it is invisible on the plate.

• Protein: a palm-sized portion of chicken, fish or beef is roughly 110-180 kcal before cooking fat.
• Starch: a cupped hand of rice, pasta, potato or bread sits around 150-220 kcal — restaurant portions are frequently two or three of these.
• Vegetables: essentially free unless they're glazed, roasted in oil, or fried.
• Added fat: +120 kcal per tablespoon, and assume at least one for anything cooked in a pan.
• Drinks: the one people forget entirely. A glass of wine is around 125 kcal, a pint of beer 180-220, a large flavoured coffee can pass 300.

How wrong will my estimate be — and does that ruin the whole thing?

Wrong by a lot, and no. Self-reported food logging averages somewhere in the range of 20-50% underreporting even for motivated, experienced trackers under laboratory scrutiny — that's true at home with a scale, not just at a restaurant. If perfect accuracy were the requirement, nobody would ever have succeeded at this.

What actually makes tracking work is consistency of bias. If you always underestimate restaurant meals by a similar amount, your log is still a reliable relative signal: you can see that this week ran higher than last week, and adjust. A log with a stable 20% error is enormously more useful than a log with six missing days. We went through this in detail in the calorie counting accuracy myth — the precision people chase is not the variable that determines whether they get results.

Should I log before or after I eat?

Before, if the meal is one you can pre-decide, and there's decent evidence this changes what you order rather than merely recording it. But the realistic answer for a restaurant is immediately after, at the table or on the walk out. Memory for portion size decays fast and asymmetrically — by the next morning, the bread basket and the second glass have quietly evaporated from your recollection while the salad remains vivid.

This is the single highest-leverage habit change in this entire article. Not better estimates. Faster logging. The gap between logging at the table and logging tomorrow is far larger than the gap between a careful estimate and a sloppy one.

What about buffets, shared plates and tapas?

Shared food defeats portion estimation because you never see your own total in one place. The workaround is to log as you go rather than reconstructing at the end: each time something lands on your plate, add it. Three quick additions during the meal beat one impossible act of archaeology afterwards.

For a buffet, most people find a flat placeholder more honest than a component breakdown — around 1,000-1,300 calories for a full plate-plus-seconds, more with dessert and alcohol. It's crude. It is also far closer to the truth than the 600 you'd arrive at by adding up the items you happen to remember.

Do I need to log alcohol separately?

Yes, and it's worth its own line rather than being folded into the meal. Alcohol carries 7 calories per gram — closer to fat than to carbohydrate — and it arrives in a form that produces no satiety at all. It also tends to correlate with everything else on the table going up. Keeping it as its own entry is what lets you notice, three weeks later, that the pattern isn't restaurants in general but specifically the nights with a second bottle.

Should I eat less earlier in the day to bank calories?

Modestly, yes; aggressively, no. Trimming a few hundred calories at lunch to make room for dinner is straightforward calorie banking and works fine. Skipping meals entirely to "save up" tends to backfire: you arrive ravenous, order more, eat faster, and stop noticing fullness — the mechanism we unpacked in why you're always hungry in a deficit. A normal breakfast and a lighter lunch beats fasting until 8pm.

How do restaurant meals fit into a weekly picture?

This is the frame that makes the whole thing tractable. Nobody eats a daily target; people eat a weekly average with a very lumpy distribution. Two restaurant meals adding 700 calories each over your normal day amounts to about 1,400 extra calories a week — meaningful, entirely manageable, and completely invisible if you stop logging on exactly those days. Our piece on the weekend calorie gap covers what that lumpiness looks like when you actually measure it.

Which brings the whole article back to one point: the value of logging a restaurant meal is not the number. It's that the day exists in your record at all.

What if I genuinely have no idea?

Then log the description and move on. "Thai green curry, rice, one beer" with a rough guess attached is a real data point. You can refine the estimate later if you care; more likely you won't need to, because what you'll want in three weeks is the pattern, not the arithmetic. The failure mode of food tracking has never been imprecision. It's abandonment — a subject we covered in why you quit tracking apps, and the answer is always the same: reduce the friction of the hardest logging moment, because that's the one that ends the streak.

Frequently asked questions