Is MyFitnessPal accurate? Where its numbers come from

Is MyFitnessPal accurate? For calories, mostly yes, when the right entries are picked. In published studies that checked its food data against research databases, daily calorie totals usually landed within about 5 percent.

Protein and fat sometimes came out a little low. Sodium and cholesterol were often well off.

The database isn’t the only source of error. A 2019 study of new users found they left out almost a fifth of food items. Here is where MyFitnessPal’s numbers come from, what the research found, and how to make your own log more trustworthy.

Key takeaways

MyFitnessPal’s database mixes entries the company added or reviewed (marked with a check) and entries members submitted, which it has not reviewed.

In studies against research databases, its daily calorie totals were often within about 5 percent of the reference. One small test of a single 3-day record was about 11 percent low.

Sodium and cholesterol are its weak spots, often because entries leave those fields empty. Fiber ran about 4 g a day low in one study.

Missed food items can cost accuracy too. Logging everything matters as much as picking the right entry.

Nutrition labels are estimates as well. US rules treat a food as misbranded only if its tested calories run more than 20 percent above the label.

Where do MyFitnessPal’s numbers come from?

MyFitnessPal is a database tracker: you search for a food, pick an entry and set a portion. As of October 30, 2026, its home page says it has “over 18 million global foods,” including packaged goods, restaurant items and recipes you save yourself. Its own help pages give 19 million, so treat the count as approximate.

Those entries come from two places. MyFitnessPal’s help article on where its food data comes from (read October 30, 2026) says the database combines “foods added by MyFitnessPal” and “foods that are added by our users.”

The check mark tells them apart. Per its help page on check-marked items, a check means an entry “has been reviewed or added by MyFitnessPal and is considered accurate and complete.” The same page adds that “even verified entries can sometimes have mistakes.”

In a March 2025 post on how its food database works, the company says a food without a check was submitted by a member and “has not been reviewed.”

That is the honest picture: a huge database, partly crowd-sourced, with a review mark on some entries. The size helps you find almost anything. The crowd-sourcing means the same food can show up many times with different numbers.

Is MyFitnessPal accurate for calories?

Usually close enough to be useful. Four peer-reviewed studies compared MyFitnessPal’s calorie numbers with a reference. All four tested its food data.

Coached users in Belgium

A 2020 study in the Journal of Medical Internet Research (PMID 33084583) had 50 adults keep two 4-day food records in MyFitnessPal. The researchers compared them with Belgium’s food composition database.

The average day came to 1,984 calories in the app against 1,958 in the reference, with a strong correlation (r=0.96).

Those participants had help. They got an illustrated manual, were asked to weigh food, and used generic foods the researchers had entered themselves. During data cleaning, the researchers removed values from 79 of 2,826 food items (2.8 percent). They say their strong results were partly due to the manual, which everyday users don’t have.

Recalls entered by a researcher

A 2018 study in Public Health Nutrition (PMID 29534771) had a trained researcher enter thirty 24-hour food recalls into five free apps. The reference was a US research database (NDSR).

MyFitnessPal averaged 1,626 calories a day against 1,696. That is about 4 percent low, and the difference was not statistically significant.

Weighed meals

A 2025 trial in the Journal of Nutrition (PMID 41022156) gave 44 US adults pre-weighed foods and had them log for three days after training. Per the abstract, MyFitnessPal’s calorie estimates were statistically equivalent to the weighed intake, within the study’s bounds of plus or minus 18 percent. The mean percent error was 3.7 percent.

One 3-day record

A 2023 study in JMIR Formative Research (PMID 37327038) entered a single person’s 3-day food record into five apps. MyFitnessPal came to 4,224 calories against 4,770 in NDSR, about 11 percent low.

It is one record, so it is weak evidence. Branded items matched closely: three Oreos came to about 160 calories in both.

Sources: Evenepoel and colleagues, J Med Internet Res, 2020 (daily mean, after data cleaning); Griffiths and colleagues, Public Health Nutr, 2018 (daily mean); Gioia and colleagues, JMIR Form Res, 2023 (3-day total). Percentages calculated from each paper’s reported means.

What it means for you: if you choose entries with care, MyFitnessPal’s calorie totals are close enough to follow a trend. For a side-by-side with other trackers, see Cronometer vs MyFitnessPal and Lose It vs MyFitnessPal.

Where is MyFitnessPal less accurate?

Beyond calories, the gaps get bigger. Three patterns show up across the studies.

Sodium and cholesterol

In the 2020 Belgian study, MyFitnessPal logged 1,293 mg of sodium a day against 2,638 mg in the reference. It logged 55 mg of cholesterol against 242 mg.

The authors concluded the app was accurate for calories, macros, sugar and fiber “but not for cholesterol and sodium.” They traced it to entries that “only have a value for total energy content” and nothing else. Fiber still ran about 4 g a day low.

Protein and fat, sometimes

In the 2018 study, MyFitnessPal’s protein (69 g against 75 g) and fat (61 g against 71 g) came out significantly low over full days. Calories did not, so the gap showed up in what the calories were made of.

Duplicate entries that disagree

The Belgian researchers entered generic foods themselves because, in their words, “generic items often have multiple entries in MyFitnessPal with highly variable nutritional information.”

A 2024 study in JMIR mHealth and uHealth (PMID 39233588) checked 42 foods high in saturated fat or cholesterol. In the English version of MyFitnessPal, cholesterol was missing for 32 percent of the foods checked. Only about half of the entries named a data source.

What you track	What the studies found	What to do
Calories	Usually within about 5% of a reference	Fine for following a trend
Protein and fat	Sometimes a little low over a full day	Prefer check-marked or label-matched entries
Fiber	About 4 g a day low in one study	Treat as a rough guide
Sodium and cholesterol	Often far off, frequently missing	Don’t rely on the app for these

Missed entries can matter as much as wrong ones

A good database can’t count a snack you didn’t enter. A 2019 study in Nutrition (PMID 30184514) asked 43 Australian adults who were new to MyFitnessPal to keep 4-day records in it.

Per the abstract, they left out an average of 18 percent of food items. Compared with two 24-hour recall interviews, the app records underestimated energy by 1,863 kJ a day, about 445 calories.

Most found the app easy, yet only 20 percent said they would keep using it. They named matching foods, judging portions and the time logging takes.

This study measured something the database studies above could not, but together they suggest a missed entry can matter as much as a wrong one. That fits a point we make often: the tracker you keep using, fully, beats the one with the cleanest data.

Where photo trackers struggle with oil and sauces, database trackers struggle with whatever you forget to type. Our guide to how AI calorie counters work covers the photo side.

Even a nutrition label is an estimate

It helps to know that the reference point isn’t exact either. US rules are in 21 CFR 101.9, current as of October 2026. Calories on a label are rounded to the nearest 5 up to 50, and to the nearest 10 above that.

The same rule treats a food as misbranded if its tested calories are “greater than 20 percent in excess” of the label. Coming in under the label is treated more loosely. So a 200 calorie snack can contain somewhat more than 200 and still be compliant.

None of this makes tracking pointless. Every tracker gives an estimate, and steady estimates still show a trend. If you track macros too, our guide on how to track macros without weighing everything covers the 4, 4 and 9 math.

How to get more accurate numbers out of MyFitnessPal

You don’t need a lab. Five habits cover most of the gaps the studies found:

Prefer check-marked entries. MyFitnessPal says these were reviewed or added by the company. They are not perfect, but they start from a better place.
Match packaged food to its label. If two entries disagree, pick the one that matches the package in your hand, or scan it if you have that feature.
Watch for empty fields. An entry with calories but no protein, fat or sodium is incomplete. Skip it if you track those.
Log oil, drinks and bites. These are easy to forget, and forgotten items were one reason the 2019 study’s records ran low.
Report bad entries. MyFitnessPal’s help pages (as of October 2026) describe a “Report Food” option in the app, and editing foods on the website.

Then watch the direction over a few weeks, not any single day. If your logged average and your weight trend keep disagreeing, your log may be running low. Talk to a doctor or a registered dietitian if you want personal guidance.

Tracking is a poor fit for some people

If counting food brings on stress, anxiety or restriction, or you have a history of disordered eating, tracking may do more harm than good. Talk with a doctor or a registered dietitian first.

Where Getfiit fits

We make Getfiit, an AI nutrition and fitness coach in WhatsApp. Instead of searching a database, you send a photo, a voice note or a sentence, and it replies with calories, protein, carbs and fat.

It is an estimate too. When a photo or message leaves something open, like oil or butter, it asks before it logs. The exchange below is an illustration, not a measured result:

lunch was a turkey sandwich and an apple, plus a latte

Got it. Was the latte made with whole milk or something else?

oat milk, medium

Logged. If anything looks off, reply to the entry or tap Edit to fix it.

There is no new app to install, because it runs in WhatsApp. As of October 30, 2026, it costs $19.99 a month after a 3-day free trial, and you can cancel anytime. See how it works on the home page.

If you are weighing other trackers too, our list of MyFitnessPal alternatives covers what each offers.

Frequently asked questions

How accurate is MyFitnessPal for calories?

In studies against research databases, its daily calorie totals were often within about 5 percent of the reference. One 2025 trial with weighed food found a mean error of 3.7 percent. A single 3-day record in a 2023 study was about 11 percent low, so results vary from study to study.

Are MyFitnessPal’s green check marks accurate?

They are entries MyFitnessPal reviewed or added, and a better starting point than member-submitted ones. MyFitnessPal’s own help page says even check-marked foods “may have nutrition information inaccuracies,” so a quick look at the label still helps.

Is MyFitnessPal accurate for calories burned?

The studies above tested food data, not exercise. If your burned figure comes from a watch or fitness tracker, that device’s estimate is what you see. A 2020 review in the British Journal of Sports Medicine (PMID 30194221) found the accuracy of wrist and arm devices varied by activity type, so treat the figure as a rough estimate.

Is MyFitnessPal more accurate than a photo calorie app?

They make different errors. A database tracker relies on picking the right entry and logging everything. A photo app relies on what the camera can see. For the photo side, see whether Cal AI is accurate. More answers are in the FAQ on the home page.

Your next step

Open your log from yesterday and look at three entries. Is each one check-marked or matched to a label, and did you log the oil and the drinks? Fix what’s off, and keep going.

This article is general information, not medical advice. Talk to a doctor or a registered dietitian before changing how you eat, especially if you have a health condition, are pregnant, or have a history of disordered eating.