How accurate are TDEE calculators?

1. The formula

For a 30-year-old man at 5 ft 9 in and 176 lb, moderately active:

Mifflin-St Jeor and Harris-Benedict usually sit within 5% of each other. Formulas built on lean mass swing more, because a body-fat guess of 20% versus 30% changes lean mass by 17.6 lb. In the systematic review by Frankenfield and colleagues, Mifflin-St Jeor predicted measured resting metabolism within 10% for more people than any other equation, which is why this site uses it by default and shows the others for comparison.

2. The activity multipliers

Most sites use 1.2, 1.375, 1.55, 1.725 and 1.9. Some insert a 1.465 step, some run a different scale from 1.4 to 2.4, and some use a slider. The same "moderately active" man gets 2,710 at 1.55, 2,562 at 1.465 and 2,404 at 1.375. Then there is how you read the labels: "3 to 5 days a week" sends most gym-goers one level above where their whole day actually lands. This is the largest source of error by far, and it is yours to control; the activity level guide is about getting it right.

3. Units and rounding

176 lb is 79.8 kg and 5 ft 9 in is 175.3 cm. A site that rounds to 80 kg and 175 cm computes a BMR of 1,749 instead of 1,749, a 0 kcal difference before the multiplier. Sites that round the multiplier, or round BMR before multiplying, add a few more. Individually these are small; stacked, they explain the last 30 to 60 kcal between two otherwise identical calculators.

4. What the calculator silently assumes

Every equation was fitted to a population of healthy adults, and the systematic review that compared them noted that older adults and US ethnic minorities were underrepresented both in building and in validating the equations. If you are very muscular, very lean, over 65, have a thyroid condition, or take medication that changes energy use, the population average applies to you less well. The equation does not know any of that, and it does not know whether you fidget, sleep cold, or stand at your desk, which is why non-exercise activity varies between people by hundreds of calories a day.

How accurate is a TDEE calculator, then?

For most healthy adults, the resting part (BMR) is within ±10% about four times out of five, and the activity part adds error on top of that. Treat the result as a range, which is how this site shows it, and treat the first two weeks as a test. Your weight trend against your logged intake is the only measurement you can make at home, and the tracker turns it into a number.

Wearables are not the answer either: in a validation across seven wrist devices, none estimated energy expenditure with less than 20% error, even though heart rate was accurate. A watch is good at counting steps and poor at counting calories.