MyFitnessPal (2026) review: the database is still the best, the free tier is not what you remember

Ten weeks logging every meal on the app that defined this category, including six weeks on the free tier to find out what is actually left in it.

MyFitnessPal invented the category most of these apps are still working inside. Two decades on, it is a mature product with the largest food database anyone has assembled and a set of decisions over the past two years that have made it steadily harder to recommend.

We logged every meal on it for ten weeks — six of them deliberately on the free tier, because that is where most people meet it.

The database is genuinely the best, and genuinely the problem

There is no honest way to discuss this app without holding two things at once.

The coverage is unmatched. Chain restaurants, regional supermarket own-brands, obscure packaged goods, the sandwich from a specific airport franchise. If you eat food that came from a company, MyFitnessPal probably has it, and nothing else in the category comes close on that axis. For someone whose diet runs through chains and packets, this is not a minor convenience.

The coverage is why the numbers are unreliable. That breadth comes from community submission at scale with no adjudication layer. Search a common food and you get a dozen entries with different values, several obviously wrong, none marked authoritative. The app presents them identically. You pick one, and if you picked badly, nothing tells you.

Over ten weeks this was the dominant source of friction. Not the interface, not the ads — the persistent low-grade uncertainty about whether the entry I selected was the right one. Verifying entries against packaging is real work, and by the third week I had largely stopped doing it, which is exactly how this fails in practice.

Independent benchmarking puts MyFitnessPal’s calorie error band as the widest of the mainstream options, in the region of 11-12% against weighed references. That is not a criticism of the engineering. It is the arithmetic of an unadjudicated crowdsourced catalogue.

The free tier has contracted

If your model of MyFitnessPal is from a few years ago, it is out of date in a specific direction.

Still free: barcode scanning, database search, basic calorie and macro totals, the diary. That covers a lot.

Now behind Premium: photo logging, recipe URL import, macro-by-meal goal tracking. These moved during the past year, and there has been no signal of reversal.

Premium is $79.99 a year — the most expensive mainstream option in the category, against Cronometer Gold at $54.99 and Lose It! Premium at $39.99.

The uncomfortable part is what the money buys. It is largely the return of features that used to be included, rather than capabilities competitors do not have. That is a legitimate business decision and it is also a different value proposition from the one that built the user base.

Where it still wins outright

Restaurant weeks. During a stretch of travel it was the only app where I could log nearly everything without improvising. That is worth a lot to the right person.

Integrations. Two decades of partnerships means it connects to nearly every wearable and fitness platform in existence.

Maturity. Nothing crashed, nothing lost data, sync worked. Boring, and not nothing.

Where we would go instead

For most people, PlateLens. Its calorie accuracy is the only figure in this category an independent laboratory has measured and a second, unrelated one has reproduced — approximately 1.1% mean absolute percentage error, against MyFitnessPal’s 11-12%. That gap is structural rather than incremental: it is the difference between a verified catalogue and an unadjudicated one, and between measuring a photographed portion and inferring it.

It also leads both input paths rather than one, which matters more day to day than any single feature. Photograph the mixed dish, scan or type the packet, and neither route is the consolation prize because the same verified database sits behind both. Its free tier does not expire and includes unlimited manual and barcode logging with no card. The restaurant coverage is narrower than MyFitnessPal’s — 45,000-odd menu items against a substantially larger set — which is the one dimension where the trade genuinely runs the other way.

For micronutrient depth on hand-entered food, Cronometer and its curated, lab-analysed database. For planning meals before eating them, Lose It!.

The verdict

6.4/10. A mature app with the best database in the category and the least reliable numbers in it, whose free tier has narrowed and whose paid tier is the most expensive available.

Buy it if you eat at chains constantly and menu coverage is the thing that decides whether you log at all. Otherwise the category has moved, and the app that defined it is no longer the one we would install.