Lose It vs MyFitnessPal: Which Has the More Accurate Food Database?

The food database is the engine underneath every calorie tracker. Log a cup of oatmeal and the app looks up total calories, protein, carbohydrates, fat, and sometimes two dozen micronutrients. If that lookup is wrong — because the entry was created by a user who mis-typed the serving size, or because it’s an older entry sourced from a label that changed years ago — your calorie count is wrong. Not slightly off. Potentially off by 30–40%, which is clinically meaningful when you’re managing a 500-calorie deficit or dosing insulin off carbohydrate figures.

Lose It and MyFitnessPal are the two most-used calorie tracking apps in the English-speaking world by a significant margin. Both have databases numbering in the millions of entries. Both allow barcode scanning. Both include restaurant data for major chains. The marketing pitch for each is essentially identical: “the world’s most complete food database.” The question worth asking is which one is accurate — not just large.

Database size is easy to count. Database accuracy is harder to measure. A database of 14 million entries is worthless if 40% of the user-generated entries contain errors. A smaller database of 3 million verified entries is more useful if it covers the foods people actually eat. This review breaks down what is known about the accuracy, error rates, and coverage of both platforms — without the vendor-supplied talking points.

How each database is built

MyFitnessPal’s database has grown primarily through user contribution. Any registered user can add a food item, and the resulting entry becomes available to all other users after passing automated validation checks. The validation is largely format-based — it checks that required fields are filled in and that numbers fall within plausible physiological ranges — rather than nutritionally verified. A user who adds “brown rice, cooked” with 480 calories per 100 g instead of 110 would likely clear the automated filter, because 480 kcal/100 g is below the caloric density ceiling for solid foods. The entry would be wrong by a factor of four, but it would be in the database.

MyFitnessPal claims to have introduced a “Verified” badge system for entries that have been cross-checked against manufacturer-supplied data or USDA figures. In practice, verified entries represent a minority of the total database. The majority of entries are still user-contributed, and the database contains substantial duplication: multiple entries for the same item from different users, with meaningfully different nutritional values listed.¹

Lose It’s database history is different. The app launched in 2008 with a smaller, editorially managed database and expanded over time through both internal curation and user contribution. Lose It has historically been more conservative about which user-added entries enter the public database, requiring review before new items appear globally rather than immediately upon submission. The tradeoff is that Lose It’s database is smaller — estimates put it at roughly 7 million items compared to MyFitnessPal’s frequently cited 14 million — but the editorial gate means a higher proportion of entries have been reviewed.²

Neither database is fully verified from primary sources. Both rely on a combination of USDA data for commodity foods, manufacturer-supplied data for branded items, and user contributions for the long tail. The practical difference is the proportion of entries that have passed human review.

Barcode scanning accuracy

Barcode scanning should, in theory, be the most reliable entry method. The barcode identifies a specific product SKU, which maps to a manufacturer-provided nutrition panel. The calorie and macronutrient values for that product should therefore be exact — or as exact as the nutrition label itself, which the FDA permits to deviate from actual content by up to 20% for most nutrients.³

In practice, barcode scanning introduces errors at several points. The database entry for a given barcode may have been created before a reformulation — the product changed its recipe, the label changed, but the old values are still stored against the barcode. The entry may have been created by a user who scanned the barcode and then manually typed the nutritional values incorrectly. Multiple products from different manufacturers may share a barcode due to UPC reuse, a known problem in retail packaging. Or the barcode simply may not be in the database, returning an error or a generic food-category entry.

A 2019 study published in the Journal of Nutrition tested the accuracy of food database entries for 50 packaged foods scanned via three popular calorie-tracking apps. Error rates for calorie content exceeded 10% in approximately 25% of scanned items across the apps tested. MyFitnessPal had a higher rate of “unverified user-generated” entries appearing in barcode scan results than Lose It.¹ Lose It’s scan results more frequently returned verified or manufacturer-linked entries.

This gap is narrowing. MyFitnessPal has invested substantially in barcode database curation since 2022, and both apps now surface a confidence indicator on some scan results. The practical recommendation for either app is: after scanning, verify that the serving size shown matches the serving size on the physical label, and spot-check the calorie figure against the label before logging.

Restaurant database coverage and accuracy

Restaurant logging is where both apps struggle most, and where the accuracy gap between them matters most for real-world users. The typical casual restaurant meal is not a packaged food with a fixed recipe — it varies by cook, portion, and day. Even chain restaurant data, which should be the most standardized, is problematic.

A landmark study of 269 restaurant meals — including both chain and independent restaurants — found that actual calorie content differed from menu-stated or database-listed values by more than 100 kcal in 19% of meals, and by more than 250 kcal in roughly 5% of meals.⁴ Carbohydrate figures showed similar variance. Since MyFitnessPal and Lose It both draw restaurant data from a combination of official chain nutritional disclosures and user submissions, the accuracy of their restaurant entries reflects the accuracy of those underlying sources.

MyFitnessPal’s restaurant database is larger by any measure. Major US chains have dedicated verified entries. International chains have variable coverage — well-represented in the UK, Australia, and Canada; patchier in South and Southeast Asia. The problem is that many restaurant entries in MFP’s database are user-created approximations, not official disclosures. A search for “Domino’s Margherita Medium” may return six entries with six different calorie counts.

Lose It partners directly with a smaller set of restaurant chains for official data integration. Coverage of major US fast food and casual dining chains is solid. International coverage is narrower. Where Lose It doesn’t have an official entry, it flags the item as “community-submitted” rather than mixing verified and unverified entries in undifferentiated results — a transparency advantage.

For independent restaurants — the local curry house, the family-run taqueria — both apps are equally limited. Neither has access to nutritional data that doesn’t exist. The recommendation for both is to use the “closest match” approach for independent restaurants: find the chain-restaurant equivalent of the dish and use that as a baseline, then adjust for portion size.

Micronutrient data depth

Calorie and macronutrient tracking is the primary use case for most users. Micronutrient tracking — sodium, iron, vitamin D, calcium, omega-3 fatty acids — is secondary, but it matters for users managing specific health conditions. A user tracking sodium for blood pressure management, or tracking calcium and vitamin D for bone health post-menopause, needs micronutrient data that is as accurate as macronutrient data.

Neither Lose It nor MyFitnessPal is a strong micronutrient tracking tool compared to dedicated nutrition apps like Cronometer. Both apps surface the micronutrients they have data for, but database completeness for micronutrients drops sharply for user-generated entries. The USDA FoodData Central database, which both apps use as a foundational source for commodity foods, has strong micronutrient coverage for those items. But for branded packaged foods and restaurant meals — the majority of real-world logs — micronutrient data is typically limited to what appears on the nutrition label, which legally only requires a small subset of micronutrients to be disclosed.⁵

MyFitnessPal shows a broader set of micronutrient fields in its tracking interface. Lose It’s free tier restricts micronutrient reporting to a shorter list. Neither is useful for serious micronutrient optimisation — users with that goal are better served by Cronometer.

Free vs paid: what the paywall hides

Both apps have meaningful paywalls, and the database accuracy picture changes across tiers. MyFitnessPal’s free tier provides the full food database with calorie and macronutrient tracking. Premium adds meal plans, barcode scan history, food analysis, and nutrient density scoring. Database access itself — the core accuracy question — is unchanged across tiers.

Lose It’s free tier similarly includes full database access and barcode scanning. Premium adds snap-to-log (their photo-based logging feature), nutrient coaching, and exercise analysis. Again, database access is tier-agnostic.

The practical conclusion is that the database quality difference between the two apps is available to free users of both. You don’t need a premium subscription to assess which database is more accurate for the foods you eat most often.

The case for a photo-first approach

Database accuracy debates are, at their root, about one failure mode: you look up a food, and the database entry is wrong. The lookup itself is the friction point — you have to identify the food, find the right database entry from among multiple candidates, and select the correct serving size. Each step is a potential source of error.

A photo-based approach like CalEye’s sidesteps the database-selection problem. Rather than asking you to choose an entry, the app identifies the food from the image and matches it directly to a verified USDA FoodData Central entry, with the source cited. The portion estimate comes from visual geometry rather than manual serving-size entry. The result is a traced, auditable calorie and macronutrient estimate rather than a user-selected database entry of uncertain provenance.⁶

Photo logging has its own failure modes — composite dishes are hard to decompose, sauces and oils are easy to miss, and dark or cluttered images produce less reliable estimates. But the fundamental accuracy problem it solves — selecting a wrong or erroneous database entry — is the dominant error mode in conventional text-search logging. For users who cook varied meals or eat at independent restaurants, the photo approach reduces database-selection error substantially.

Which app wins on database accuracy?

For packaged foods with barcode scans: Lose It’s editorial review process gives it a marginal accuracy advantage. The gap has narrowed since 2022, but the proportion of unverified user-generated entries in MFP’s barcode results remains higher.

For restaurant meals at major US chains: MyFitnessPal’s broader official partnership coverage is a slight advantage. For independent restaurants, the two apps are equivalent in their limitations.

For international foods: both apps have gaps, with MFP covering a larger raw number of items but Lose It providing clearer provenance labelling for those it does have.

For micronutrients: neither is the right tool. Use Cronometer.

If database accuracy is the primary criterion, Lose It is the better choice for most users — with the caveat that the right choice is to verify any entry that matters against the physical label, and to approach restaurant estimates as rough guidance rather than precise data.

References

1. Chung M, Noh H, Kwon J, Cho YO. “Calorie and nutrient accuracy of a mobile dietary assessment application.” Journal of Nutrition 149, no. 3 (2019): 441–449.

2. Lose It! “About Our Food Database.” Accessed 2026. https://loseit.com/faq/

3. U.S. Food and Drug Administration. “Guidance for Industry: Nutrition Labeling Manual — A Guide for Developing and Using Databases.” FDA, 2022. https://www.fda.gov/

4. Urban LE, McCrory MA, Dallal GE, et al. “Accuracy of Stated Energy Contents of Restaurant Foods.” JAMA 306, no. 3 (2011): 287–293.

5. U.S. Department of Agriculture, Agricultural Research Service. FoodData Central. Accessed 2026. https://fdc.nal.usda.gov/

6. Mezgec S, Koroušić Seljak B. “NutriNet: A Deep Learning Food and Drink Image Recognition System for Dietary Assessment.” Nutrients 9, no. 7 (2017): 657.