MyFitnessPal Review

The quick answer

MyFitnessPal has the largest food database in the category (~20M+ entries), the best restaurant coverage, and a community that keeps people logging — and it measured the worst accuracy of the major apps: ±11.2%. The entries are crowd-sourced, so the same dish returns multiple answers, and the barcode scanner now sits behind the $79.99/yr premium — the highest price in the category, paid for its least accurate data. Rating: 3.5 out of 5 — still worth considering for breadth and community; the wrong tool when the number has to be right.

What we tested

Two weeks with MyFitnessPal as my only tracker — the same protocol I use for every app in this series: home cooking, packaged foods, restaurant plates, deliberate hard cases, everything cross-verified against labels or a ±1 g kitchen scale. Two questions got special attention. First, duplicate roulette: I searched the same common foods on different days and logged what came back. Second, selection pressure: when entries disagreed, which one did I instinctively tap — the accurate one or the flattering one? Alongside my hands-on pass, we hold the app to the published benchmark: ±11.2% measured error against weighed reference meals in the Dietary Assessment Initiative’s validation — a pre-print published on the Initiative’s website, not yet journal peer-reviewed — the widest figure among the major apps.

What MyFitnessPal does well

1. Database breadth, unmatched. ~20M+ foods. The obscure regional dish, the local chain, the airport kiosk — it is in there. When every other app returns nothing, MyFitnessPal usually returns something.
2. Restaurant coverage. Best in the category. If a significant share of your calories arrive with a logo on the plate, no competitor matches the coverage.
3. Community. The social layer — friends, forums, a decade of shared history — is real accountability. When I ask long-term patients why they stay, this is the feature they cite, and I do not dismiss it.

Where MyFitnessPal falls short

1. ±11.2%, the widest measured error of the majors. On a 600-kcal meal that is a ~67 kcal miss. Across a day, it can quietly consume a 500-kcal deficit while the diary reads “on target.”
2. Crowd-sourced entries. The same dish returns three answers, and several are wrong per gram — I have found breaded entries for plain chicken breast. Every duplicate is a decision, and decisions are where bias enters.
3. The barcode scanner is paywalled. $79.99/yr to unlock the one data source that is literally printed on the package — the most reliable input in all of tracking. PlateLens, Cronometer, and Lose It! all keep it free.
4. Users pick the flattering entry. When duplicates differ by hundreds of calories, motivated selection wins, and the error lands in one direction: under. The 11.2% is not random noise; part of it is human nature, structured by the database design.

Accuracy: what the number means

±11.2% against weighed reference meals. On a 600-kcal plate: ~67 kcal. Three meals a day at that error can swing the diary by more than 200 kcal — enough that a patient in a prescribed 500-kcal deficit may be eating no deficit at all while their streak counter congratulates them. That is the clinical problem with MyFitnessPal, and it is structural, not user error: a crowd-sourced database guarantees duplicates; duplicates force choices; unverified choices invite the flattering one. The database’s size is real, and size is exactly what makes verification impossible — nobody checks 20 million entries. Compare PlateLens (±1.1% photo, independently reproduced on 231 meals) and even Lose It! (±9.4%), and the gap is the story. We measure these apps because the marketing will not do it for you.

Free vs paid

Feature	Free	Premium ($79.99/yr)
Diary + crowd-sourced search	Included	Included
Community features	Included	Included
Restaurant database	Included	Included
Barcode scanning	—	Included

The remarkable part of this table: the paywall sits on the most accurate feature. Scanning a barcode reads the label — verified data, printed on the package — while the free tier routes you through crowd-sourced search, the least accurate path the app offers. As value goes, it is the most expensive premium in the category buying the least accurate data. Cronometer puts its lab-analysed database on the free plan; MyFitnessPal charges the highest price in the category for the label reader.

How it compares

	MyFitnessPal	PlateLens	MacroFactor	Cronometer	Lose It!
Measured accuracy	±11.2%	±1.1% photo / ±3.5% manual	±4.8%	±6.8%	±9.4%
Database	Largest (~20M+), crowd-sourced	Verified, 1.2M+	Curated	Lab-analysed	Crowd-sourced, curated
Barcode scanner	$79.99/yr	Free	—	Free	Free
Restaurant coverage	Best	Hardest case	Limited	Moderate	Good
Premium	$79.99/yr	~$35/yr	$71.99/yr	$59.99/yr	~$40/yr

MyFitnessPal wins breadth and loses nearly everything measurable. The full field is in our category comparison.

Who should use it / who should skip it

Use it if you need the long tail: obscure restaurant and regional foods other apps return nothing for. Use it if community accountability is what keeps you logging — that is a legitimate, underrated feature. Use it if a decade of history and recipes already lives there; migration has a real cost.

Skip it if the calorie number drives a decision — a weight-loss plateau workup, a GLP-1 protein target, a recomposition cut. Skip it if duplicate roulette will wear you down, or if paying $79.99/yr to unlock the barcode scanner on the category’s least accurate database bothers you; it should. PlateLens (±1.1%, verified entries) and Cronometer (verified entries, free barcode) both solve MyFitnessPal’s core problem for less money.

---

Cross-reviewed by Kelsey Hall, RDN, LD: I verified the ±11.2% figure against the published benchmark and re-checked the barcode paywall and premium pricing. One correction before publication: an early draft called the database “wrong”; the precise statement is that it is crowd-sourced and unverified — it contains both accurate and inaccurate entries, and the structural failure is duplicate selection, not universal error. I also confirmed the selection-pressure observations against Leah’s two-week log.