Why Your Calorie Tracking App Might Be Lying to You

Key takeaways

• Researchers tested four popular photo-based calorie-tracking apps — MyFitnessPal, LoseIt!, CalAI, and Appediet — against meals prepared and measured to the nearest 0.1 gram in a controlled NIH metabolic kitchen.
• Across all four apps, calorie estimates came in roughly 250 to 345 calories too low per meal on average, and fat was underestimated by about 30 grams — both discrepancies of roughly one-third.
• High-fat meals, including ketogenic-style dishes, were the hardest for the apps to estimate accurately, while carbohydrate estimates were more consistent across the board.

A rare, precise way to test accuracy

Most of us have no real way to know how accurate our calorie-tracking app actually is — we take a photo, get a number, and move on. This study did something most of us can't: it compared app estimates against meals whose exact nutritional content was already known down to the tenth of a gram, because they'd been built for a separate NIH nutrition study happening in a controlled metabolic kitchen. That's about as close to ground truth as food research gets, which makes this a genuinely useful stress test rather than just another consumer comparison.

Researcher Aaron Hengist noted that this kind of direct, high-quality comparison hadn't really been available before — these apps are used constantly, but rigorously checking their accuracy against precisely known meals is much rarer than you'd expect.

The gap was bigger than a rounding error

The results weren't a minor miss. Across all four apps, calorie estimates landed about 250 to 345 calories too low per meal on average — roughly one-third of the meal's actual calorie count. Fat was underestimated by a similar margin, averaging about 30 grams short per meal. Carbohydrate estimates, by contrast, were notably more consistent, suggesting the apps' image-recognition and database systems handle carb-heavy foods more reliably than fat-dense ones.

Two of the apps, MyFitnessPal and LoseIt!, were more accurate for higher-calorie meals than lower-calorie ones — worth knowing if you're tracking smaller meals or snacks throughout the day, where the margin of error may matter more relative to the total.

Why fat is the tricky one

The pattern here makes some intuitive sense. Fat is calorie-dense and often visually harder to quantify from a photo than, say, counting the amount of rice or bread on a plate. That challenge showed up most clearly in ketogenic-style meals, which tend to be higher in fat by design — these were the dishes the apps struggled with most. If a meal leans heavily on oils, dressings, nuts, cheese, or fatty cuts of protein, this research suggests there's a good chance the app is leaving out a meaningful chunk of what you actually ate.

What this means if you track your food

None of this means calorie-tracking apps are useless — they're clearly faster and more convenient than manually logging every ingredient. But this research is a useful reality check: if you're using a photo-based app without adjusting portions or occasionally cross-checking with manual entry, what you see on screen is likely an undercount, not an overcount. That's a meaningful distinction if you're using the app's output to guide decisions about intake, since the built-in bias runs in the direction of underestimating rather than overestimating what you consumed.

The researchers' practical suggestion is to treat these apps as a helpful starting point rather than a precise measurement — pairing photo-based tracking with occasional manual entry or portion-checking, especially for fat-heavy meals, to get a more realistic picture.

The takeaway

This is still early-stage, preliminary work, but the testing method was rigorous, and the size of the discrepancy is hard to ignore. If tracking accuracy matters to you, the most useful habit here isn't abandoning your app, it's building in a little healthy skepticism, especially around higher-fat meals, and occasionally spot-checking a few entries against a manual estimate to calibrate your sense of the gap.