By Samipan Bandyopadhyay, developer of Fitto AI
Restaurant meals are where most food logs go quiet. There is no label, no scale and no recipe, the dish may have no entry in any database, and the person across the table is not waiting while you search for one. So the meal gets skipped, or logged as a vague guess, and those tend to be the biggest meals of the week.
You can’t make a restaurant meal exact. What you can do is estimate it the same way every time, so that it is close enough and you keep logging. Here is a method that works whether the menu prints calories or not.
Why restaurant meals are the hardest to log
There are two problems, and they stack.
The meals are bigger than they look. Researchers at Tufts measured 364 meals from non-chain restaurants in Boston, San Francisco and Little Rock in a calorimeter. The average came to 1,205 calories per meal, and American, Italian and Chinese meals averaged about 1,500. 92% of the meals held more energy than a typical person needs at one sitting, and that is before drinks, appetizers or dessert.
People guess low, and guess lower as the meal gets bigger. In a study of diners at 89 fast-food restaurants, adults underestimated their meal by 175 calories on average, and adolescents by 259. The bigger the meal, the bigger the miss.
So an unaided guess at a restaurant meal is usually too low, and furthest off on the biggest meals. Each step below exists to counter that.
Step 1: Use the posted number when there is one
In the US, restaurants that are part of a chain with 20 or more locations must post calories on their menus and menu boards, and have done since May 2018. They must also give you written nutrition information on request, covering total fat, saturated fat, carbohydrates, sugars, fiber, protein and more. That makes a chain meal the easy case: the number is printed in front of you, and most chains publish the full breakdown on their website.
How far can you trust it? When researchers measured 269 items from 42 restaurants, the stated calories were within 10 calories per portion on average, which is close. But 19% of the items held at least 100 calories more than their label said, and it was the items listed as lower in calories that tended to hold more than stated.
The posted number is the best one you will get. Two adjustments make it better:
- Add your changes. Extra cheese, a side of dressing, a swapped side or a refill is not in the printed figure. Log each one as its own line.
- Check what the number covers. Menus often list a pizza or a shared plate per slice or per serving. Multiply by what you actually ate.
Step 2: No number? Split the plate into its parts
Most independent restaurants publish nothing. Don’t go looking for a database entry with the same dish name: “chicken pad thai” in a database is someone else’s recipe at someone else’s portion, and there are usually dozens of entries with different numbers.
Estimate the parts instead. Almost any plate breaks down into four:
- Protein. Meat, fish, tofu, eggs. A cooked portion the size of your palm is about 85 to 115 g (3 to 4 oz).
- Starch. Rice, pasta, bread, fries, tortillas. A fist is about a cup. A cup of cooked white rice is about 205 calories.
- Vegetables. Usually the smallest number on the plate, unless they are fried or coated in sauce.
- Fat and sauce. The part that hides the most calories, so it gets its own step.
Each part is easier to judge on its own than the whole dish at once. Our guide to tracking calories without weighing food has the full set of hand measures.
Step 3: Add the fat you can’t see
Restaurant kitchens cook with oil and butter, finish dishes with more, and serve sauces and dressings in generous amounts. Very little of it shows in a photo or on the plate, and per gram it is the densest food on the table. These are reference values from USDA FoodData Central. The portions are ours, so use the per-100 g column for your own amounts.
| Hidden extra | Portion | Calories per 100 g | Calories |
|---|---|---|---|
| Olive or vegetable oil | 1 tbsp (13.5 g) | 884 | 119 |
| Butter | 1 tbsp (14 g) | 717 | 102 |
| Ranch dressing | 2 tbsp (30 g) | 430 | 129 |
| Mayonnaise | 1 tbsp (14 g) | 680 | 95 |
| Sour cream | 2 tbsp (24 g) | 198 | 48 |
| Guacamole | ¼ cup (60 g) | 155 | 93 |
| Tortilla chips | 1 cup (30 g) | 472 | 142 |
A dish that looks glossy has oil on it. Seared, sautéed, stir-fried and pan-fried food usually carries at least a tablespoon per portion, so adding one oil line to those dishes is a reasonable default. If a sauce or dressing comes on the side and you use all of it, log all of it.
Step 4: Know where each cuisine hides its calories
Every cuisine has one or two items that carry most of the meal. Look for them first.
- Mexican. Chips arrive before you order, and a basket is often several cups. Then come the toppings: guacamole, sour cream and cheese, each a line of its own. A burrito’s tortilla alone can be a big share of the total.
- Sushi. The rice is most of the energy in a roll, and rolls are dense with it. Tempura, cream cheese and mayo-based sauces (“spicy” or “crunchy” rolls) add more. Sashimi is mostly fish, with no rice.
- Chinese takeout. Battered and fried dishes and sweet sauces add the most. The rice is easy to forget, and a takeout container usually holds more than a cup.
- Italian and pizza. Pasta portions are often two or more cups of cooked pasta. Bread with olive oil or butter before the meal is its own line. For pizza, count slices, not pizzas.
- Burgers and diners. Mayo and sauces on the bun, fries, and refills of soda or milkshakes. Order the burger and log the fries as a separate line.
Step 5: Log it before you eat, and round up
Take the photo or note before the first bite. At the start of the meal every part is separate and visible. Halfway through, the sauce has mixed into the rice and you can’t tell what was there.
When you’re unsure, pick the larger estimate. Diners guess low, and they guess lowest on the biggest meals, so rounding up corrects in the right direction. If the true portion is somewhere between one cup and two, log one and a half.
Log it even when the number is rough. A rough figure for a 1,200-calorie dinner tells you far more than a gap in the log. If restaurant meals keep getting skipped, your weekly averages make every week look lighter than it was. A logging streak that doesn’t break on a heavy day helps here. You get credit for logging the big meal, not for avoiding it.
How to log a restaurant meal in Fitto AI
Photograph the plate before you eat, or say or type what you ordered. Fitto AI lists each food it can identify on its own line, with an estimated portion and calories, and it doesn’t need the dish to be in a database. A local special works the same way as a chain’s menu item. Before anything is saved you can edit every line, change a portion, or add the oil, butter or dressing the camera couldn’t see. A photo, voice or text analysis costs 1 AI credit, and new accounts get 5 free.
Three things make eating out easier to keep logging:
- Regular orders are one tap. Star a meal you order often as a favourite, and re-logging it costs no credit.
- Packaged drinks and snacks scan for free. Use the barcode for anything with a label.
- Ask for a lighter version. After a scan you can ask for lighter versions of the same meal to try next time, such as dressing on the side or a different side dish.
If you want to know how far to trust a photo estimate, we wrote up what AI photo calorie counters get right and wrong. Restaurant food, with its hidden fat, is the hardest case.
Fitto AI provides estimates for general wellness. If tracking food ever starts to feel stressful or compulsive, it’s worth talking to a doctor or registered dietitian.
Sources
- Urban LE, Weber JL, Heyman MB, et al. Energy contents of frequently ordered restaurant meals and comparison with human energy requirements and US Department of Agriculture database information: a multisite randomized study. Journal of the Academy of Nutrition and Dietetics, 2016;116(4):590–598.
- Block JP, Condon SK, Kleinman K, et al. Consumers’ estimation of calorie content at fast food restaurants: cross sectional observational study. BMJ, 2013;346:f2907.
- Urban LE, McCrory MA, Dallal GE, et al. Accuracy of stated energy contents of restaurant foods. JAMA, 2011;306(3):287–293.
- U.S. Food and Drug Administration. Menu Labeling Requirements. Retrieved October 2026.
- U.S. Department of Agriculture, Agricultural Research Service. FoodData Central, SR Legacy: oil, olive, salad or cooking (FDC ID 171413); butter, salted (173410); salad dressing, ranch, regular (173592); mayonnaise, regular (171009); cream, sour, cultured (171257); tortilla chips, plain, white corn, salted (167558); rice, white, long-grain, enriched, cooked (168878). FNDDS: guacamole (2709307). Values per 100 g, retrieved October 2026.
Related reading
- Best AI Calorie Tracker Apps for Android (2026)
- How to Count Calories in Homemade Food and Recipes
- How Accurate Are AI Photo Calorie Counters?
More from Fitto
- Everything Fitto AI does: photo, voice, text and barcode logging
- Tracking calories for restaurant and takeout meals
- A calorie tracker without a subscription
- How Fitto compares with other calorie tracker apps