Formulas estimate resting calorie burn within about 10% for most adults
High confidence
Resting metabolic rate is the calories your body burns just to stay alive at rest. The Mifflin-St Jeor formula lands within 10 percent of a lab measurement for about 4 in 5 healthy adults. It is less reliable for people carrying a lot of extra weight, for very muscular people, and for athletes, where only about half land within 10 percent.
Why it works
Most of the energy your body burns at rest comes from lean tissue (muscle and organs), not fat. The formulas estimate that lean tissue from four things you can measure: height, weight, age, and sex. From there they predict the calories it burns. Inside those cells, the mitochondria (the parts that turn fuel into energy) burn at a steady rate. So resting burn tracks closely with how much lean tissue you carry.
The evidence
-
Of the four most commonly used equations, Mifflin-St Jeor was the most reliable, predicting RMR within 10% of measured in more nonobese and obese individuals than any other equation, with the narrowest error range. Noteworthy errors remain when it is applied to individuals and possibly to some age and ethnic groups. No individual-level validation was found for WHO/FAO/UNU.
-
Mifflin-St Jeor was unbiased and landed within 10% of measured RMR for 82% of people, similar to Livingston (79%); the other equations tended to overestimate. Accuracy was lower in obese than non-obese people for every equation (87% vs 75% for Mifflin-St Jeor).
n=337
-
This retrospective study compared basal metabolic rate (the calories the body burns at rest) across 133 overweight and obese adults using indirect calorimetry (breath-based measurement), bioelectrical impedance, and two prediction equations. Measured BMR by indirect calorimetry averaged 1,581 kcal/day, which was significantly lower than the 1,766 kcal/day estimated by bioelectrical impedance and the 1,788 kcal/day from Harris-Benedict. Of the equations tested, the Mifflin-St Jeor formula tracked the measured values most closely in this population, though it still overestimated (1,690 kcal/day) and landed within 10% of the measured value for only about half of people (50.4%).
n=133
-
The authors built and tested a revised Harris-Benedict equation for estimating resting metabolic rate using data from 722 adults (173 men, 549 women), with resting metabolic rate measured by indirect calorimetry. The new equations had R-squared values of 0.95 for men and 0.86 for women and showed the lowest bias and root-mean-square error of the equations compared (Harris-Benedict, Mifflin-St Jeor, Owen, FAO/WHO/UNU). The margin over the original Harris-Benedict was modest: in the held-out test group of 40 men and 127 women, predictions landed within 10% of measured rate for 67.5% of men and 59.1% of women, versus 65% and 57.5% for the original, with mean bias of 8.32% vs 9.06% (men) and 8.93% vs 9.49% (women).
n=722
-
This study tested how well eight common equations predicted resting metabolic rate against breath-based indirect calorimetry in 30 healthy adults aged 18 to 65 with body mass index from 19 to 39. At the group level the Harris-Benedict and World Health Organization equations were most accurate, with small average bias of -14 and -25 kcal/day. But none of the equations performed well for individuals, and as fat-free mass (muscle and organ mass) increased, the equations increasingly underestimated resting metabolic rate.
n=30
-
This study tested how well resting-energy-expenditure equations held up in overweight and obese class I and II adults (body mass index 25 to 40) in separate US and Dutch samples. In the US group the Mifflin-St Jeor equation was the most accurate, predicting within 10% of measured value for 79% of people with almost no average bias (-1.0%). In the Dutch group no equation reached that level of accuracy, so the study concluded there was no reliably accurate equation for Dutch overweight and obese adults.
n=546
-
This is the 1990 paper that created the Mifflin-St Jeor equation. Researchers measured resting energy expenditure by indirect calorimetry in 498 healthy adults (247 women, 251 men, ages 19 to 78), including both normal-weight and obese people, then used regression to link it to weight, height, age, and sex. The resulting equation explained 71% of the variation in resting energy (R-squared 0.71): 10 x weight in kg plus 6.25 x height in cm minus 5 x age in years, plus 5 for men or minus 161 for women.
n=498
Why we call confidence high
A systematic review of the four common equations (Frankenfield 2005) found Mifflin-St Jeor the most reliable, landing within 10% of measured RMR in more nonobese and obese people than any other, with noteworthy individual errors. A 337-adult validation (Frankenfield 2013) found it unbiased and within 10% for 82% of people (87% non-obese, 75% obese), and Weijs 2008 found 79% in US overweight adults. Original and revised Harris-Benedict do somewhat worse (about 6 in 10 within 10% in Pavlidou 2023), no equation was accurate in Weijs's Dutch sample, and in athletes Mifflin-St Jeor lands within 10% for only about half (O'Neill 2023 meta-analysis).
Where it applies
Healthy, non-athlete adults aged roughly 18-65, mostly white North American and European samples; accuracy is lower in people with obesity.
Does not apply to: athletes and highly trained runners (equations land within 10% for only about half; O'Neill 2023); people with high fat-free mass (equations increasingly underestimate RMR; Flack 2016); older adults and ethnic groups underrepresented in validation studies.
Last reviewed Oct 14, 2026. See how we score.