CaRoot
JPEN

How far off are your calorie logs? What measured intake shows

Most people assume their food log is roughly right. When researchers check self-reported calories against an objective measurement, they find the opposite: reported intake runs low, and the gap is often large.

In a pooled analysis of five large validation studies, people under-reported calories by an average of 15% on a single 24-hour recall and 28% on a food frequency questionnaire. That sounds like bad news for anyone keeping a log. But the same body of research also links regular logging with more weight loss. Both can be true, and the difference changes how you should read your numbers.

How researchers know what you really ate

The reference method is doubly labeled water. A participant drinks water tagged with stable isotopes, and over the next couple of weeks urine samples show how many calories their body burned. For someone whose weight is stable, calories burned and calories eaten are about the same, so researchers can compare that figure with what the person reported.

A second check exists for protein. Urinary nitrogen reflects how much protein someone actually ate, which gives researchers a separate measure that doesn't rely on memory.

The size of the gap

Study

Method

Calories under-reported

OPEN study, 2003 (men)

24-hour recall

12–14%

OPEN study, 2003 (women)

24-hour recall

16–20%

OPEN study, 2003 (men)

Food frequency questionnaire

31–36%

OPEN study, 2003 (women)

Food frequency questionnaire

34–38%

Freedman et al., 2014 (5 studies pooled)

Single 24-hour recall

15% on average

Freedman et al., 2014 (5 studies pooled)

Food frequency questionnaire

28% on average

Neither method is quite the same as a daily food log. OPEN didn't test food records at all. A 2009 review that did cover diet records and recalls found that about 30% of people were under-reporters, and that energy intake was underestimated by roughly 15%. A 2019 systematic review of 59 studies and 6,298 adults came to a similar place: most studies found under-reporting, and 24-hour recalls showed less of it than other methods.

Under-reporting also isn't spread evenly. The pooled analysis found that BMI strongly predicted it. A 2025 analysis in Nature Food, built on 6,497 doubly labeled water measurements, flagged 27.4% of dietary reports in two national surveys as implausible, with the gap widening as BMI rose.

The most striking number comes from a small 1992 study in the New England Journal of Medicine. It looked at people with obesity who said they couldn't lose weight despite eating little. They had under-reported their food intake by about 47% and over-reported their physical activity by about 51%. That was a selected group, so it doesn't describe a typical logger, but it does show how far perception can drift from measurement.

Protein holds up slightly better

In OPEN, protein was under-reported by 11–12% for men and 11–15% for women on 24-hour recalls. On those recalls, that's a little less than calories. So your protein number is also likely to be low, just not by as much.

Why the gap appears

Studies are better at measuring the gap than explaining it. One documented cause is reactivity: when people know their intake is being recorded, what they eat can change. Models using traits such as BMI and social desirability did predict some under-reporting in OPEN, but they explained relatively little of it.

From a practical standpoint, the usual suspects are the items that are easy to miss: cooking oil, a second pour of something, bites while cooking, drinks, sauces. Treat that as a checklist rather than a research finding. We didn't find a study that puts a reliable number on each one.

One day tells you very little

Even a perfect log has a second problem: your intake moves around from day to day.

A classic study by Basiotis and colleagues followed 29 people who recorded their food every day for a year. To estimate one person's usual calorie intake to within 10%, it took an average of 31 days of records. Estimating the average for a group took only 3 days. A later analysis by Nelson and colleagues reached a related conclusion: seven days of records don't rank people as accurately as researchers commonly assumed.

That's the case for reading averages over weeks instead of judging a single day. One high day is mostly noise, and so is one low day.

Inaccurate logs still seem to help

A 2011 systematic review of 22 studies found a consistent link between dietary self-monitoring and weight loss. The authors rated the evidence as weak, since the samples were narrow and the data were themselves self-reported. So this is an association, not proof that logging causes weight loss.

A 2019 study adds a useful detail. In a 24-week online weight management program with 142 participants, people spent an average of 23.2 minutes a day on self-monitoring in the first month and 14.6 minutes by the sixth. Time spent wasn't what separated those who lost more. Frequency was: people who lost at least 10% of their weight opened their food diary about 2.7 times a day, compared with 1.7 for those who didn't.

We didn't find a study that directly tests whether trends in a log stay useful despite the error, so the next section is reasoning, not a research result.

How to read your own numbers

Treat your log as a consistent gauge rather than a precise scale. If you tend to miss the same kinds of things each week, your total will be off, but this week's number can still be compared with last week's.

  • Use averages. Look at a week or more, not a single day.
  • Calibrate against your scale. If your log says you're in a clear deficit but your average weight hasn't moved in a few weeks, your log is probably running low. Adjust your target based on what the scale shows.
  • Check the easy misses. Oils, drinks, and sauces are worth logging even as a rough guess.
  • Keep it quick. In the study above, frequency mattered more than minutes. A rough entry made every day beats a perfect one you stop making.

Some error comes from the numbers themselves, not from you. Packaged food has an allowed margin on its label, and AI photo estimates have their own limits.

Where CaRoot fits

CaRoot is a food log built around shorter entries. You can photograph a meal and let AI estimate calories and macros, scan a barcode, or search a food database, and you can edit the result before saving. The weekly report shows your average calories, macro balance, and how consistently you logged, so you're reading the week rather than the day.

It won't remove the gap described above. Estimates can be wrong, and anything you don't log won't be counted. What it can do is make logging quick enough to keep doing.

Sources

This article is general nutrition information, not medical advice. If you have an eating disorder, or a health condition that affects what you should eat, talk to your doctor or a registered dietitian. Calorie tracking may not be right for everyone.

CaRoot logs a meal from a photo,
so tracking stays quick.

See the app