How much do calorie calculators actually disagree?

Two profiles were run through four published resting-metabolism equations, then through the five activity multipliers that almost every calculator uses. Nothing here was taken from another website. Each figure comes from applying the equation as published, so the whole table can be reproduced from the four papers cited at the end and a calculator.

The profiles are 32 years old, 175 cm, 70 kg, male, and 32 years old, 165 cm, 62 kg, female. The Cunningham equation predicts from lean mass rather than total weight, so it needs a body-fat percentage. It is reported at three plausible values instead of one, because assuming a single figure would hide exactly the kind of choice this analysis is about.

For the male profile, every published equation lands within 170 kcal of every other, a spread of about 10%.

The female profile spreads wider, 329 kcal or about 25%, and almost all of that comes from Cunningham. That is not a defect in Cunningham. It is the equation correctly reporting that lean mass matters, and that assuming a body-fat percentage for somebody is a bigger guess than most calculators admit.

Hold the equation fixed at Mifflin-St Jeor and change only the activity level, and the same man ranges from 1,967 to 3,114 kcal.

That is a 1,147 kcal spread from one dropdown, against 170 kcal from the entire disagreement between four equations developed across a century. The activity multiplier moves the answer roughly seven times more than the science does.

Combine both sources of variation and the same person can be told anything from 1,966 to 3,437 kcal a day. That is the range a genuinely confused user is looking at when they compare four websites.

Stop comparing calculators. Two sites differing by 200 kcal are not disagreeing in any meaningful way, and picking the one whose number you prefer is choosing an activity level by wishful thinking.

Spend the effort on the input that actually moves the result. Being honest about your activity level is worth more than any choice of equation, which is why it is worth reading how to choose your activity level before trusting any output.

Then treat whatever number you get as a hypothesis. Eat near it for two to three weeks, watch the weekly average, and let the scale tell you your real maintenance level. That figure is measured on you, which no equation can be.

This compares equations, not branded websites. A given calculator may round differently, offer a different number of activity levels, or apply a multiplier this analysis did not test, so its output can sit outside these ranges.

Two profiles are not a population. The spreads reported here describe these two people, and the gaps between equations widen or narrow at different ages, weights and body compositions.

Most importantly, none of these equations is the truth. A 2005 systematic review found Mifflin-St Jeor predicted resting metabolic rate within 10% of measured values more often than the other commonly used equations, and with the narrowest error range, which is why this site uses it. The same review was explicit that noteworthy individual errors remain, and that older adults and several ethnic groups were underrepresented in the studies these equations were built from. An equation being the best available is not the same as it being right about you.