Guide
Harris-Benedict vs Mifflin-St Jeor: Formulas and Accuracy
The revised Harris-Benedict (1984) and Mifflin-St Jeor (1990) equations both estimate resting energy expenditure from weight, height, age and sex. In validation studies against indirect calorimetry, Mifflin-St Jeor has more often landed within 10% of measured values in non-obese and obese adults, but results differ by population and either equation can be off by several hundred kcal per day for an individual. Treat the output as a starting estimate and adjust it using 2-3 weeks of tracked intake and body weight.
What the two equations calculate
Both equations are regression formulas that predict resting energy expenditure (REE), also called basal or resting metabolic rate (BMR or RMR), in kcal per day. They use four inputs: body weight (W, in kg), height (H, in cm), age (A, in years) and sex. Neither includes body composition, so two people with identical inputs receive the same estimate even if their lean mass differs.
The Harris-Benedict equations were derived from indirect calorimetry data collected in the early twentieth century. In 1984, Roza and Shizgal re-evaluated them using the original 239 normal subjects plus 98 additional subjects from Benedict's later data, and reported that the equations estimate the resting energy expenditure of a normal subject with a precision of about ±14%. The revised coefficients are listed in the table below.
The Mifflin-St Jeor equation was published in 1990. It was derived from 498 healthy subjects (247 women and 251 men, aged 19-78 years; 264 normal-weight and 234 obese) whose REE was measured by indirect calorimetry. The authors reported that the original Harris-Benedict equations overestimated measured REE by 5% in that sample.
| Equation | Men | Women |
|---|---|---|
| Revised Harris-Benedict (Roza & Shizgal, 1984) | 88.362 + 13.397 × W + 4.799 × H − 5.677 × A | 447.593 + 9.247 × W + 3.098 × H − 4.330 × A |
| Mifflin-St Jeor (1990) | 10 × W + 6.25 × H − 5 × A + 5 | 10 × W + 6.25 × H − 5 × A − 161 |
What validation studies report
A 2005 systematic review, conducted with the American Dietetic Association's evidence analysis methods, compared four widely used equations: Harris-Benedict, Mifflin-St Jeor, Owen and WHO/FAO/UNU. It concluded that Mifflin-St Jeor predicted resting metabolic rate within 10% of the measured value in more non-obese and obese individuals than any other equation and had the narrowest error range, while also reporting noteworthy errors and limitations when the equation is applied to individuals.
A 2013 validation in 337 community-living adults found the Mifflin-St Jeor equation unbiased (95% confidence interval -26 to +8 kcal per day), with 82% of estimates within 10% of measured values. Harris-Benedict tended to overestimate. Accuracy was lower in obese than in non-obese participants (75% versus 87% for Mifflin-St Jeor).
Results are not uniform across populations. In a Brazilian sample of 82 sedentary adults aged 30-60 with overweight or obesity, both equations underestimated measured REE, and Mifflin-St Jeor underestimated by more than Harris-Benedict (the paper printed the original, unrevised Harris-Benedict coefficients). In a 2023 Italian study of older adults with severe obesity, the same pattern appeared in the validation group of 382 people. The table summarizes these results.
The scale of the error matters more than the ranking. For a resting estimate of 1,500 kcal per day, an error of 10% is 150 kcal per day, and Roza and Shizgal's precision of ±14% corresponds to roughly ±210 kcal per day. The evidence supports a modest conclusion: Mifflin-St Jeor has the stronger record in the two analyses by Frankenfield, it is not uniformly more accurate in every population, and agreement at group level does not guarantee accuracy for one person.
| Study | Population | Harris-Benedict | Mifflin-St Jeor |
|---|---|---|---|
| Mifflin et al., 1990 | 498 healthy adults | Original equations overestimated measured REE by 5% | Derived from this sample (R² = 0.71) |
| Frankenfield, 2013 | 337 community-living adults | Tended to overestimate | Unbiased (95% CI -26 to +8 kcal/day); 82% within 10% |
| de Oliveira et al., 2011 | 82 adults with overweight or obesity, Brazil | -177.8 kcal/day (-8.9%) | -289.1 kcal/day (-15.6%) |
| Danielewicz et al., 2023 | 382 older adults with severe obesity (validation group), Italy | Bias -1.9%; 60% within ±10% | Bias -8.5%; 54% within ±10% |
From a resting estimate to daily calorie needs
Total daily energy expenditure (TDEE) is commonly estimated by multiplying the resting estimate by an activity factor. The five factors below are conventional values used by many calculators. The labels attached to each level vary between sources, and the sources reviewed for this article did not trace the set to a single validation study. They also differ in scale from the physical activity level (PAL) categories in the FAO/WHO/UNU energy requirements report, which lists 1.40-1.69 for sedentary or light activity lifestyles, 1.70-1.99 for active or moderately active lifestyles and 2.00-2.40 for vigorous lifestyles. The choice of activity level therefore adds uncertainty on top of the equation error.
| Activity level (labels vary) | Multiplier | TDEE with revised Harris-Benedict (kcal) | TDEE with Mifflin-St Jeor (kcal) |
|---|---|---|---|
| Sedentary (little or no exercise) | 1.2 | 1,790 | 1,742 |
| Lightly active (light exercise, 1-3 days a week) | 1.375 | 2,051 | 1,996 |
| Moderately active (moderate exercise, 3-5 days a week) | 1.55 | 2,312 | 2,250 |
| Very active (hard exercise, 6-7 days a week) | 1.725 | 2,573 | 2,504 |
| Extra active (very hard exercise or a physical job) | 1.9 | 2,834 | 2,758 |
Using the result: start, track, adjust
In the example above, the revised Harris-Benedict resting value is about 40 kcal per day (2.8%) higher than the Mifflin-St Jeor value, which becomes 48-76 kcal per day after multiplying by the activity factor. In four sample profiles calculated for this article (ages 30-50, weights 60-95 kg), the revised Harris-Benedict value was 2.8-6.3% higher at rest. That gap is usually smaller than the 250-260 kcal per day step between adjacent activity levels in the table.
Because individual error can reach several hundred kcal per day, the output is treated as a hypothesis to test. A common approach is:
- Set intake near the estimated TDEE, adjusted for the goal.
- Log intake and weigh at consistent times and conditions for 2-3 weeks.
- Compare the weight trend with what the estimate predicted. Daily readings are dominated by water and glycogen fluctuation, so the trend is more informative than any single day.
- Change the target by the observed difference and repeat. The related guide on the 7,700 kcal per kg rule explains how weight change is converted to kcal, and why that conversion is also an approximation.
Who the equations are not designed for
Both equations were derived from, and validated mainly in, generally healthy adults. The 2005 review reported that older adults and US-residing ethnic minorities were underrepresented in both development and validation studies, and advised a high level of suspicion about accuracy in groups that are greatly underrepresented. Roza and Shizgal found the Harris-Benedict equations unreliable in malnourished patients, where predicted values underestimated measurements. Additional limits to keep in mind:
- Children and adolescents, pregnancy and lactation, acute illness and conditions that alter metabolism were not examined in the sources cited here.
- Neither equation uses body composition. Mifflin and colleagues reported that fat-free mass explained more of the variation in REE than any other single variable in their sample (R² = 0.64), so people with unusually high or low lean mass may deviate from the estimate.
- Adults with severe obesity and older adults showed larger errors in the studies above, and the Danielewicz authors recommend indirect calorimetry where available.
- An activity multiplier chosen by feel is a judgment, and a one-step error moves the estimate by about 250-260 kcal per day in the example above.
Where these equations appear on this site
The calorie calculator page on this site (/calorie-calculator/) offers both equations so the results can be compared. Talerz, an iPhone calorie counter, uses the revised Harris-Benedict equation for its plan.
This article is general information and not medical advice; a qualified health professional can assess individual energy needs.
Frequently asked questions
Which is more accurate, Harris-Benedict or Mifflin-St Jeor?
In a 2005 systematic review and a 2013 validation study of 337 adults, Mifflin-St Jeor predicted resting metabolic rate within 10% of measured values for more people than Harris-Benedict, which tended to overestimate. Two other studies in adults with overweight or obesity found Mifflin-St Jeor underestimated by more than Harris-Benedict. Neither is accurate for every individual.
What are the formulas?
Revised Harris-Benedict, men: 88.362 + 13.397 × weight (kg) + 4.799 × height (cm) − 5.677 × age (years). Women: 447.593 + 9.247 × weight + 3.098 × height − 4.330 × age. Mifflin-St Jeor: 10 × weight (kg) + 6.25 × height (cm) − 5 × age (years), plus 5 for men or minus 161 for women. Results are in kcal per day.
How large is the typical error?
Roza and Shizgal reported a precision of about ±14% for Harris-Benedict in normal subjects. In a 2013 study, 82% of Mifflin-St Jeor estimates fell within 10% of measured resting metabolic rate, meaning roughly one in five did not, and the share was lower (75%) in obese participants.
Why multiply by 1.2 to 1.9?
The resting estimate covers only energy used at rest. Activity multipliers are a convention for approximating total daily energy expenditure, and the choice of level is a judgment that adds uncertainty. Tracking weight for 2-3 weeks is the practical way to check the result.
Sources
- Mifflin MD, St Jeor ST, et al. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr 1990 — Mifflin-St Jeor coefficients, derivation sample of 498 subjects, and the 5% overestimate by the original Harris-Benedict equations in that sample.
- Roza AM, Shizgal HM. The Harris Benedict equation reevaluated. Am J Clin Nutr 1984 — Re-evaluation of Harris-Benedict data, precision of about ±14%, and unreliability in malnourished patients.
- Harris & Benedict formula of Roza and Shizgal, 1984 (Table 5, Dutch guideline database) — Printed revised Harris-Benedict coefficients for men and women.
- Frankenfield D, Roth-Yousey L, Compher C. Comparison of predictive equations for resting metabolic rate in healthy nonobese and obese adults: a systematic review. J Am Diet Assoc 2005 — Systematic review ranking Mifflin-St Jeor by share of estimates within 10%, with caveats for individuals and underrepresented groups.
- Frankenfield DC. Bias and accuracy of resting metabolic rate equations in non-obese and obese adults. Clin Nutr 2013 — 337 adults; Mifflin-St Jeor unbiased with 82% within 10%; lower accuracy in obesity; other equations tended to overestimate.
- de Oliveira EP, et al. Comparison of predictive equations for resting energy expenditure in overweight and obese adults. J Obes 2011 — 82 Brazilian adults; mean differences for original Harris-Benedict and Mifflin-St Jeor versus indirect calorimetry.
- Danielewicz AL, et al. Prediction of resting energy expenditure in Italian older adults with severe obesity. Front Endocrinol 2023 — Validation group of 382; relative bias and share within ±10% for Harris-Benedict and Mifflin-St Jeor.
- FAO/WHO/UNU. Human energy requirements: report of a joint expert consultation (Section 5, Table 5.3) — Physical activity level categories of 1.40-1.69, 1.70-1.99 and 2.00-2.40.