Why dietitians switched from Harris-Benedict to Mifflin-St Jeor

In 1919 Francis Benedict and James Harris published the first predictive equation for resting metabolic rate (RMR). It was state-of-the-art science: 239 healthy adults measured by direct calorimetry. For 70 years, dietitians and exercise physiologists used it without a serious competitor.

Then in 1990 a team led by Mifflin and St Jeor published a re-derivation in 498 modern adults. Their conclusion was understated: the old equation overestimates RMR by about 5%. That sounds small until you scale it up — for a 90 kg adult cutting 500 kcal/day to lose weight, a 5% error is roughly 100 kcal, which is the difference between losing weight and stalling.

This article is what I wish I’d read before I ever opened a calorie calculator. It walks through:

• What the head-to-head studies actually measured
• Where Harris-Benedict misses (and where it’s still fine)
• How big the practical difference is for everyday people
• When to ignore both equations and just measure

If you came here to plug numbers into a tool, the BMR & TDEE calculator on this site uses the Mifflin-St Jeor equation by default. Read on to understand why.

What “RMR” actually means

Resting metabolic rate is the energy your body uses lying still, fasted, awake, in a thermoneutral environment. It’s roughly 60–75% of total daily energy expenditure (TDEE) in a sedentary adult, falling to about 50% in a serious athlete.

There are two ways to know it: measure it (indirect calorimetry — a metabolic cart at a hospital or sports lab) or estimate it (an equation that takes weight, height, age, and sex). The accuracy of the equation only matters if measurement isn’t an option, which for most people most of the time is the case.

The studies that broke Harris-Benedict

The decisive evidence came from a series of papers between 1990 and 2005:

Mifflin et al., 1990 (PMID 2305711). The original derivation. 498 adults across a wide BMI range had RMR measured by indirect calorimetry. The Mifflin equation explained 71% of the variance versus 65% for Harris-Benedict, and its error was symmetrical — Harris-Benedict was systematically high.

Frankenfield, Rowe, Smith & Cooney, 2003 (PMID 14600563). 130 hospitalised adults. Mifflin-St Jeor was within 10% of measured RMR in 80% of cases versus 64% for Harris-Benedict. Harris-Benedict’s error widened with BMI — by far the most clinically important finding because the people most likely to be using a calorie calculator are people working on body composition.

Frankenfield, Roth-Yousey & Compher, 2005 (PMID 15883556). The systematic review that ended the debate. Pooling data from multiple studies, Mifflin-St Jeor was the most accurate of the four mainstream equations across both nonobese and obese populations. The Academy of Nutrition and Dietetics adopted it as the predictive equation of choice the same year.

How big is the difference for a real person?

The gap widens with age and BMI. For a 30-year-old of healthy weight, the practical difference is small (~5%). For an older adult with obesity, Harris-Benedict can overshoot by 8% or more — easily 200 kcal/day. That’s the difference between a sustainable deficit and chronic frustration.

Where Harris-Benedict is still fine

If you’re using an older calculator that hasn’t been updated since 2010, your number will be 50–150 kcal high. For maintenance, that doesn’t matter — your body will absorb the noise. For a deliberate cut or gain, it does matter, because the whole point of the calorie target is to bias the energy balance in a known direction. Use a Mifflin-St Jeor calculator when the deficit/surplus is the design.

The 1984 Roza-Shizgal revision of Harris-Benedict is a reasonable middle ground if you happen to be looking at it. It cleaned up some of the historic problems with the 1919 equation but didn’t fully close the gap with Mifflin in obese populations.

When to ignore both equations

Three situations call for measurement instead of estimation:

• Hospitalised or critically ill patients. Both formulas underperform when basal metabolism is altered by illness, sepsis, or trauma. ESPEN guidelines recommend indirect calorimetry as the reference method for clinical nutrition decisions (ESPEN 2017).
• After very large weight loss. Resting metabolic rate adapts; equations don’t know that.
• High-level athletes with unusual body composition. Equations are derived from average humans. A 100 kg athlete with 8% body fat is not average.

For everyone else — the vast majority of people who type “TDEE” into a search bar — Mifflin-St Jeor is good enough, and our BMR calculator and macro calculator both anchor on it for that reason.

Practical takeaway

Don’t agonise over which equation to use. Pick the modern one (Mifflin-St Jeor), set a realistic activity multiplier, and adjust based on two weeks of actual scale data. The equation isn’t the bottleneck — adherence and consistency are.

If you want a number to start from now, the BMR & TDEE calculator takes 30 seconds and uses Mifflin-St Jeor by default. If you’re working on body composition, the macro calculator builds on the same equation and adds protein/carb/fat splits per ISSN 2017 guidance.