The formulas, and why calculators disagree

Every calorie calculator starts with a formula for your resting burn, then multiplies it by an activity level. The four common formulas usually land within about 150 calories of each other. One step on the activity scale moves the answer by 250 to 350, so the word you pick matters more than the formula.

Why do calculators give you different numbers?

You put the same height, weight and age into three sites and get three answers, sometimes 500 calories apart or more. It looks like nobody knows anything. Mostly the sites made two different choices: which formula to use for your resting burn, and what each activity word means.

The formulas work like guessing a shoe size from someone’s height. Tall people usually have bigger feet, so you’re close most of the time, and now and then you’re two sizes off. The formulas guess how much working tissue you carry from your height, weight, age and sex. For most bodies the guess is good. For a very muscular body, or one with more fat than average for its weight, it drifts.

The four formulas, oldest first

Lean mass is your weight minus your fat: weight × (1 minus your bodyfat as a fraction). The two lean-mass formulas ignore age, sex and height, because those mostly stand in for how much lean tissue you have. In Cunningham’s 1991 review, lean mass alone explained 65 to 90% of the differences in resting burn between people.

How close do they get?

Strong Mifflin-St Jeor and Harris-Benedict have each been checked against lab measurements in hundreds of people, more than once.

A 2005 systematic review in the Journal of the American Dietetic Association compared the formulas used most in clinics. Mifflin-St Jeor came within 10% of the measured number for more people than any other, and had the narrowest spread of errors. The review also warned that older adults and US ethnic minorities were underrepresented in the studies.

A 2013 test by the same lead author measured 337 adults living ordinary lives. Mifflin-St Jeor was unbiased: its average error sat within 26 calories a day of zero. Most of the others ran high. Mifflin-St Jeor came within 10% for 82% of people, 87% of those without obesity and 75% of those with it.

The 1984 revision of Harris-Benedict described its own precision as about 14% either way. When Mifflin’s team tested the old equations on their 498 adults, Harris-Benedict overestimated by about 5% on average.

So the formula is close for most people and more than 10% off for about one in five. On a resting burn of 1,600 calories, 10% is 160 a day. Multiply through an activity level and it’s 200 to 300 calories of maintenance, about half a typical deficit.

Two people, four formulas

Take a 35 year old woman, 5 foot 5 and 170 pounds, with about 35% bodyfat, who walks 6,000 steps a day. And a 40 year old man, 5 foot 10 and 200 pounds, about 28% bodyfat, who walks 4,000 steps and lifts three hours a week.

Cunningham sits highest for both. At the same lean mass it runs 130 calories plus 0.4 per lean kilogram above Katch-McArdle, about 150 for these two, and that’s the whole spread.

Now hold the formula still and change the activity word. Here is the woman’s Mifflin-St Jeor number on the usual calculator scale.

Each step is about 257 calories, close to twice the gap between her formulas. For the man the gap is starker. His steps put him at sedentary, and his lifting adds about 100 calories a day across the week, so his maintenance is about 2,290. A calculator with “moderately active” ticked, because he goes to the gym three times a week, prints 2,826. That’s 540 calories a day of error, and none of it came from the formula.

When does knowing your bodyfat help?

Mifflin-St Jeor assumes you’re built like the average person of your height and weight. Two 200 pound men of the same height and age get the same answer from it, 1,823 calories. If one is at 15% bodyfat and the other at 35%, Katch-McArdle gives them 2,035 and 1,644. That’s almost 400 calories a day apart at rest.

So bodyfat helps most at the edges: very muscular people, whom the height and weight formulas undercount, and people with more fat than average for their weight, whom they overcount. It only helps if the bodyfat number is good. In a 2023 test of 15 bioimpedance devices on 73 adults, the error for a single measurement ran from 3.3 to 12.6 points of bodyfat depending on the device. For a 200 pound man, 5 points of bodyfat is 10 pounds of lean mass, about 100 calories a day in Katch-McArdle. We’d only plug in a number from a lab measurement, such as a DEXA scan.

Why age is in some formulas and not others

Mifflin-St Jeor takes 5 calories a year off for age, and the revised Harris-Benedict about 4 to 6. The lean-mass formulas take nothing off. The 1984 revision found resting burn followed body cell mass, the working tissue, and didn’t depend on age or sex once that was counted. A 2021 study of 6,421 people agrees from a different angle: adjusted for body size and composition, daily burn held steady from 20 to 60. Age shows up in the height and weight formulas mostly because people tend to lose lean tissue as they get older, not because the years themselves cost calories. More in metabolism and age.