How accurate are calorie calculators?

Published 13 September 2026  ·  Guides  ·  Calso

Accurate enough to start from, not accurate enough to trust to the last few hundred calories. The formula inside most calculators, Mifflin-St Jeor, estimated resting metabolic rate within 10% of the measured value for 82% of 337 adults in a 2013 validation study, and for 75% of those with obesity. The bigger error comes after the formula: for the 82 kg man in the example below, each step up the activity menu adds about 308 kcal a day.

Below is where each part of the number comes from, how far each part can be off, and a three-week method that replaces the calculator's guess with a figure from your own scale and your own log.

How does a calorie calculator work out your number?

Almost every calculator does the same three things, and each one adds its own error:

  1. Estimate your resting burn from sex, age, height and weight, usually with the Mifflin-St Jeor equation.
  2. Multiply it by an activity factor chosen from a menu, from about 1.2 for "sedentary" to about 1.9 for "extra active".
  3. Add or subtract for your goal, usually by assuming a fixed number of calories per kilogram or pound of body weight.

Here is that arithmetic for one person, a 35-year-old man, 178 cm tall and 82 kg, who picks "lightly active" and a pace of 0.5 kg a week. It is the calculation Calso's target starts from, which we describe further down.

Step Calculation Result
Resting burn, Mifflin-St Jeor 10 × 82 + 6.25 × 178 − 5 × 35 + 5 1,763 kcal
Maintenance, lightly active 1,763 × 1.375 2,423 kcal
Target, losing 0.5 kg a week 2,423 − (0.5 × 7,700 ÷ 7) 1,873 kcal

How accurate is the Mifflin-St Jeor formula?

For most adults it lands within 10% of a laboratory measurement, and for nearly one person in five it does not. It is also the best of the common options. A 2005 systematic review in the Journal of the American Dietetic Association looked at the four equations most used in clinical practice and found Mifflin-St Jeor predicted resting metabolic rate within 10% for more people than any other, with the narrowest error range.

A 2013 study in Clinical Nutrition then measured 337 community-living adults by indirect calorimetry and tested seven equations against them:

On our example, a 10% miss on 1,763 kcal is 176 kcal at rest, and 242 kcal once the activity factor multiplies it. The 2005 review adds a warning worth reading twice: older adults and ethnic minorities living in the US were underrepresented both in building these equations and in testing them.

Why does the activity level matter more than the formula?

Because the multiplier is a single guess from a menu, and each step on that menu moves the answer by more than the formula's typical error. Applied to our example's 1,763 kcal:

Activity level Multiplier Maintenance
Sedentary 1.2 2,115 kcal
Lightly active 1.375 2,423 kcal
Moderately active 1.55 2,732 kcal
Very active 1.725 3,040 kcal
Extra active 1.9 3,349 kcal

There is a second problem with the bottom of that menu. The 2004 FAO, WHO and UN University report on human energy requirements classifies a sedentary or light activity lifestyle at 1.40 to 1.69 times basal metabolic rate, and states that the values free-living adult populations sustain over long periods range from about 1.40 to 2.40. A meta-analysis it cites, covering 411 adults aged 18 to 64, found a most common value of 1.60 in a group its authors described as having a "predominantly sedentary Western lifestyle".

The comparison is approximate, because the report multiplies a basal rate measured under stricter conditions than the resting rate Mifflin-St Jeor predicts. The gap is still not small: 1.60 applied to our example gives 2,820 kcal, which is 705 kcal above the "sedentary" answer. If you have a desk job and no exercise habit, a calculator's sedentary figure is more likely to sit below what you burn than above it.

Why does weight loss slow down on the same calorie target?

Because the goal step assumes your body burns the same amount at every weight, and it does not. Most calculators convert a weekly pace into a daily cut with the 3,500 kcal per pound rule, or its metric twin of about 7,700 kcal per kilogram, and assume the loss continues in a straight line.

A 2011 paper in The Lancet, led by Kevin Hall of the US National Institutes of Health, modelled what actually happens. For a 100 kg sedentary man cutting 480 kcal a day:

So a target built from a weekly pace is right about the direction and wrong about the timetable. Expect the scale to move fastest at the start and slower each month after, and do not read the slowdown as a sign that you are logging badly.

How do you find your real maintenance calories?

Log what you eat every day for three weeks, weigh yourself every morning, then let the change in your weekly average tell you what the calculator could only guess. The arithmetic takes a minute:

  1. Log every meal for 21 days, including the days you would rather not.
  2. Weigh yourself each morning, same scale, before eating.
  3. Average the weights in week 1 and in week 3. Ignore single readings.
  4. Average your logged calories per day across the three weeks.
  5. Maintenance = average intake + (weight lost in kg × 7,700 ÷ 14). The 14 is the number of days between the middle of week 1 and the middle of week 3.

An illustrative example, not a real user: our 82 kg man picked "sedentary", so his calculator said 2,115 kcal to maintain. He logs an average of 1,950 kcal a day, and his weekly average falls from 82.0 kg to 81.3 kg.

Figure Value
Weight lost between week 1 and week 3 0.7 kg
Daily deficit, 0.7 × 7,700 ÷ 14 385 kcal
Real maintenance, 1,950 + 385 2,335 kcal
Calculator, sedentary 2,115 kcal, 220 low
Calculator, lightly active 2,423 kcal, 88 high

This uses the same 7,700 kcal figure criticised above, but Hall's objection is to projecting it across months while expenditure falls with weight. Across a fortnight and 0.7 kg, that effect has very little room to build.

The method has one more advantage. Your logged calories carry your tracker's bias, whether it comes from a photo estimate or from typing. If that bias is steady, the maintenance you calculate carries it too, and targets set against it still land where you mean them to. What breaks the method is a bias that changes from week to week, which is why we covered how accurate photo calorie counters are and why the calorie-dense foods are worth weighing even without a food scale habit.

What does Calso's daily target use?

Disclosure

We publish Calso. The values below are read from the app's own source code on 13 September 2026, so everything on this page about calculator error applies to our number too.

Calso builds your target from the nine questions in onboarding. An AI model drafts the plan starting from the formula below, and if its answer falls outside a plausible range, the app uses the formula's own number instead.

Two consequences, in plain terms. Our sedentary multiplier of 1.2 sits below the FAO range for sedentary lifestyles, so if you picked Sedentary and you are losing faster than the pace you chose, the likely fix is Lightly active. And our pace uses the static rule, so expect weekly loss to slow over the months on an unchanged target. You can change your activity level and weekly pace from your profile, and saving recalculates your targets. None of this is medical advice, as our Terms set out.

FAQ

Is Mifflin-St Jeor the most accurate calorie formula?

Of the four equations most used in clinical practice, yes. A 2005 systematic review found it predicted resting metabolic rate within 10% for more people than any other equation it reviewed, with the narrowest error range. In a 2013 study of 337 adults it was the only one of seven equations with no measurable bias, though the Livingston equation came close on accuracy, 79% against 82%.

Would knowing my body fat percentage make a calculator more accurate?

Not automatically. In the 1990 study that produced the Mifflin-St Jeor equation, fat-free mass on its own explained 64% of the variation in resting energy expenditure, while weight, height, age and sex together explained 71%. A body fat figure also has its own measurement error, and it does nothing about the activity multiplier, which is the larger source of error.

Why do two calorie calculators give me different numbers?

Because they use different formulas, different activity multipliers, or both. In the 498 people used to build Mifflin-St Jeor, the older Harris-Benedict equation overestimated measured resting expenditure by 5%. In a 2013 comparison of seven equations, five tended to overestimate. Then each calculator defines its activity levels in its own words, and one step on that menu is worth hundreds of calories.

Why has my weight loss slowed down on the same calorie target?

Because you burn less as you weigh less. A 2011 Lancet model led by a US National Institutes of Health researcher found that a fixed cut in daily intake produces weight loss that slows over time, with about half the eventual change reached in a year and 95% in about three years. The 3,500 kcal per pound rule assumes the loss stays linear, and for their example it predicted about twice the first-year loss.

Should I get my metabolic rate measured?

Only if the estimate is failing you in a way that matters clinically, and that is a conversation for a clinician. The 2005 review noted that measuring resting metabolic rate by indirect calorimetry removes the equation's error, and advised a high level of suspicion about the equations for older adults and ethnic minorities in the US, who were underrepresented in the studies behind them.

Three things to take away

A starting target from nine questions

Calso turns your body stats, activity level, goal and pace into a daily calorie and macro target, then estimates each meal from a photo against it. The home page lists everything the app does today, and everything it does not.

Sources