Mifflin-St Jeor vs Harris-Benedict: which BMR formula should you trust?

The short answer

Use Mifflin-St Jeor. It's the formula endorsed by the Academy of Nutrition and Dietetics for predicting resting energy expenditure in healthy adults. Across multiple comparison studies — most notably Frankenfield et al. (2005) — Mifflin-St Jeor matched indirect calorimetry within 10% more often than Harris-Benedict did, and its overestimation bias was smaller. For modern body sizes and modern lifestyles, Mifflin gets you closer to the truth.

This doesn't mean Harris-Benedict is 'wrong' in any catastrophic way — it's still within striking distance for the majority of adults. But if you're choosing one formula in 2026, choose Mifflin.

The two formulas, side by side

Both formulas take age, sex, weight (kg), and height (cm) and return BMR in kcal/day. Here's how they differ:

W = weight in kg, H = height in cm, A = age in years.

Why Harris-Benedict tends to overestimate today

Harris-Benedict was built on data from 1919. The subjects were 239 individuals, predominantly male, predominantly lean, with body compositions and activity patterns that don't reflect the modern average. Body sizes have changed (we're heavier on average), body compositions have changed (lower lean mass for the same total weight), and measurement methods have improved. When you apply a formula built on early-20th-century lean bodies to a 2026 population, it tends to predict BMRs that are slightly too high.

The revised 1984 Roza-Shizgal version of Harris-Benedict updated the coefficients, but didn't fully close the gap. Frankenfield's comparison study found revised Harris-Benedict still overestimated BMR by 5–15% on average in modern adults, particularly in heavier and older subjects.

Worked comparison: same person, both formulas

Let's run the same person through both. 32-year-old male, 5'10" (178 cm), 180 lb (81.6 kg).

Mifflin-St Jeor: BMR = (10 × 81.6) + (6.25 × 178) − (5 × 32) + 5 = 816 + 1,112.5 − 160 + 5 ≈ 1,774 kcal/day

Harris-Benedict (revised 1984): BMR = 88.362 + (13.397 × 81.6) + (4.799 × 178) − (5.677 × 32) = 88.362 + 1,093.2 + 854.2 − 181.7 ≈ 1,854 kcal/day

Gap: about 80 kcal/day, or ~4.5% higher under Harris-Benedict. Multiply by an activity factor of 1.55 and the TDEE gap widens to ~124 kcal/day. That's enough to derail a moderate cut — or to keep someone from gaining weight when they think they're at a surplus.

When Mifflin-St Jeor might still be off

Even the best general-population formula has caveats. Mifflin-St Jeor will underestimate or overestimate BMR for:

• Very lean athletes with above-average muscle mass (use Katch-McArdle if you know your body fat).
• Very obese individuals at the extremes of weight (use Mifflin-St Jeor's adjusted version or work with a dietitian).
• Pregnant or breastfeeding women (caloric needs increase substantially; standard formulas aren't validated).
• People with significant thyroid dysfunction (hypothyroidism can suppress BMR by 5–10%; hyperthyroidism elevates it).
• Older adults over 80 (formula reliability decreases; case-by-case dietitian guidance is better).
• Children and teens (use the Schofield equation, designed for younger age groups).

The history, briefly

Harris-Benedict was the first widely-used BMR predictive equation, derived from 239 subjects (136 male, 103 female) studied at the Carnegie Institution between 1909 and 1917. For decades it was the standard, used in hospitals, dietetics programs, and physiology textbooks. It received a meaningful revision in 1984 (Roza & Shizgal), with updated coefficients to better fit later populations.

Mifflin-St Jeor came in 1990, derived from 498 healthy adults across a wider weight range and including more accurate indirect calorimetry. The Academy of Nutrition and Dietetics (then the ADA) compared the available formulas in a 2005 review by Frankenfield, Roth-Yousey, and Compher, and concluded that Mifflin-St Jeor was the most reliable predictor of resting metabolic rate in non-obese and obese individuals. Since then, it's been the dominant choice in clinical nutrition.

Should you ever still use Harris-Benedict?

Honestly, no — not for personal use in 2026. If you're following an older textbook or a study that used Harris-Benedict, fine, you can compute it for reference. But for setting your own diet target, Mifflin-St Jeor's slight edge in accuracy is the difference between a 100 kcal mismatch and a 200 kcal mismatch, and over a multi-month cut, that matters. Stick with Mifflin.

The 1990 study that changed which formula everyone uses

Mark Mifflin and Sachiko St Jeor weren't trying to overthrow Harris-Benedict when they ran the study that bears their names. They were trying to figure out why the existing equations seemed to consistently overestimate resting metabolic rate in their clinic patients. So they did something useful: they measured RMR via indirect calorimetry in 498 healthy adults across a wide range of BMIs, including a substantial number of overweight and obese subjects — exactly the populations Harris-Benedict had under-sampled in 1919. From that data they fit a new linear equation.

The key insight wasn't subtle. Harris-Benedict had been built on a sample dominated by lean adults (subjects in the original 1919 study averaged a BMI around 21). Mifflin-St Jeor's sample included the modern body composition spread, and the fit was cleaner. When they cross-validated the equation against held-out data, mean absolute error dropped meaningfully versus Harris-Benedict, with the gap widening for higher-BMI individuals. That finding got reproduced multiple times in the 1990s and 2000s, and by 2005 the American Dietetic Association formally recommended Mifflin-St Jeor as the predictive equation of choice for non-obese adults.

If you're curious about the original paper, it's Mifflin et al. (1990), American Journal of Clinical Nutrition. The text is sometimes available through PubMed. Most clinical nutrition reference texts now use Mifflin as the default with Katch-McArdle reserved for athletic populations.

Side-by-side BMR estimates: Harris-Benedict vs Mifflin-St Jeor

How the two equations diverge across common body profiles. Differences are usually 50–150 kcal — small at the day level, meaningful over a multi-month cut.

Mifflin reliably comes in lower than Harris-Benedict for modern bodies. The gap is largest at higher BMIs (where Harris-Benedict was poorly calibrated) and smallest for lean young adults (where Harris-Benedict's 1919 sample was most representative).

What 100 kcal of formula error costs you over a year

The Mifflin-vs-Harris gap looks small at the day level — usually 80–120 kcal. Easy to dismiss. But formulas aren't applied at the day level; they're applied as a continuous baseline that shapes months of eating decisions.

Say you're trying to maintain weight. Harris-Benedict tells you maintenance is 2,500 kcal/day. Mifflin tells you it's 2,400. You believe Harris-Benedict and eat 2,500. The reality (closer to Mifflin's number) means you're in a 100 kcal/day surplus. Over 12 months, that's 36,500 surplus calories — roughly 9–10 lb of fat gain you can't explain. People who 'eat at maintenance and still gain weight slowly' are often exactly this case. The formula was off, and a year compounds the gap.

The reverse case is worse for adherence. Harris-Benedict tells you a 20% cut is 2,000 kcal. Mifflin tells you the real maintenance is 2,400 and a 20% cut is 1,920. You eat 2,000 kcal expecting a steady deficit, the scale moves slowly, and you conclude the diet 'isn't working' — when in fact your deficit is only 400 kcal/day, not the 500 you planned for. The motivation cost of slow-feeling progress is real.

This is why the formula choice matters even though the day-level difference is small. You're not eating a formula once; you're eating it 365 times a year.

How to verify which formula matches your real body

If you want to know empirically whether Mifflin or Harris-Benedict is closer to your true metabolism, run this protocol over 14 days:

1. Calculate both Mifflin-St Jeor and Harris-Benedict for yourself. Take the midpoint as your starting estimate.
2. Multiply each BMR by your honest activity factor to get two TDEE estimates. Pick the LOWER number (likely closer to truth based on modern data).
3. Eat at exactly that calorie target for 14 days. Weigh food, log everything, no cheats.
4. Weigh yourself daily under identical morning conditions. Compute the 7-day rolling average on day 7 and day 14.
5. If the day-14 average matches day-7 average within 0.5 lb, you've found maintenance — that's your real number.
6. If the day-14 average is higher than day-7, the formula overestimated; subtract 150 kcal and retest.
7. If the day-14 average is lower than day-7, the formula underestimated; add 100 kcal and retest.

After one or two cycles, you'll know your real TDEE within ~50 kcal. At that point the formula debate becomes academic — you've verified the actual number, which is always more reliable than any prediction equation.