We Compared 5 TDEE Formulas Against 1,090 Lab Measurements — Here’s Which One Actually Works

82%. That is the percentage of non-obese adults whose resting metabolic rate the Mifflin-St Jeor equation predicts within 10% of lab-measured values. The Harris-Benedict equation? Depending on the population, it drops as low as 38%.

Most TDEE content online tells you to “use the Mifflin-St Jeor formula” without showing why. Here is the actual evidence: five prediction equations tested against indirect calorimetry — the gold standard where researchers measure your oxygen consumption and CO2 output to calculate exactly how many calories you burn at rest. We pulled the accuracy data from Frankenfield et al.’s 2005 systematic review published in the Journal of the American Dietetic Association, supplemented by a 2024 comparative analysis from Medicine (Lippincott) and the original formula publications.

The Five Formulas, Explained

Every TDEE calculator starts with a Basal Metabolic Rate (BMR) estimate, then multiplies by an activity factor. The five most common BMR equations:

The Cunningham and Katch-McArdle formulas are functionally related — both predict BMR from lean body mass. Cunningham is often cited for RMR (slightly higher than BMR) while Katch-McArdle targets BMR specifically. For practical purposes, the difference is 50-100 calories per day.

Calculate your own BMR with all these formulas using the CalcFit BMR calculator, then see your full TDEE with the TDEE calculator.

Accuracy Comparison: What the Lab Data Shows

Frankenfield, Roth-Yousey, and Compher (2005, Journal of the American Dietetic Association, n=1,090 across multiple studies) conducted a systematic review comparing predictive equations to indirect calorimetry measurements. Here is what the research actually shows:

*Cunningham and Katch-McArdle accuracy rates are from smaller validation studies and depend heavily on the accuracy of the lean body mass input. With DEXA-measured LBM, accuracy approaches 80%+ for lean individuals.

The data is clear: Mifflin-St Jeor wins for general populations. It was the most accurate across both non-obese and obese groups, and it had the narrowest error range. This is why the Academy of Nutrition and Dietetics recommends it as the default predictive equation.

Accuracy by Body Type: Where Each Formula Fails

No single formula works equally well for everyone. Here is where each one breaks down, based on Frankenfield’s data and a 2024 comparative analysis published in Medicine (Lippincott Williams & Wilkins):

Lean/Athletic Individuals (BMI 18.5-24.9, Body Fat <15% Men / <22% Women)

• Best choice: Katch-McArdle or Cunningham (if body fat is accurately measured)
• Why: These formulas use lean body mass directly, capturing the higher metabolic rate from above-average muscle mass that weight-based formulas miss
• Mifflin-St Jeor: Slightly underestimates for highly muscular individuals (by 50-150 cal/day)
• Harris-Benedict: Overestimates by 3-8% even in this group

Normal Weight, Non-Athletic (BMI 18.5-24.9)

• Best choice: Mifflin-St Jeor
• Why: 82% accuracy within 10% — the highest for this population
• Avoid: Harris-Benedict (tends to overestimate, leading to smaller deficits than intended)

Overweight (BMI 25-29.9)

• Best choice: Mifflin-St Jeor
• Why: Maintains ~75% accuracy in this range
• Harris-Benedict: Overestimates by 5-10%, making weight loss slower than expected
• Katch-McArdle: Only useful if body fat is measured via DEXA or hydrostatic weighing — skinfold and bioimpedance estimates lose accuracy above BMI 27

Obese (BMI 30+)

• Best choice: Mifflin-St Jeor (70% accuracy within 10%)
• Worst choice: Harris-Benedict (38-64% accuracy, often overestimates by 200-400 calories)
• Why formulas struggle here: Adipose tissue has lower metabolic activity than lean tissue, but weight-based formulas treat all weight equally. A 2020 study in Nutrients (Predicting Equations and Resting Energy Expenditure Changes in Overweight Adults) confirmed that prediction error increases linearly with BMI above 30.

Estimate your body fat first with the body fat calculator, then use the right formula in the TDEE calculator.

Decision Tree: Which Formula Should YOU Use?

Stop overcomplicating this. For most people, the answer is straightforward:

1. Do you know your body fat percentage from a reliable method (DEXA, hydrostatic weighing, or BodPod)?
   • Yes, and body fat is below 20% (men) or 28% (women) → Use Katch-McArdle
   • Yes, and body fat is above those thresholds → Use Mifflin-St Jeor
   • No → Go to step 2
2. Are you obese (BMI 30+)?
   • Yes → Use Mifflin-St Jeor (do not use Harris-Benedict)
   • No → Use Mifflin-St Jeor
3. Are you on GLP-1 medication (Ozempic, Wegovy, Mounjaro)?
   • Yes → See the GLP-1 section below — standard formulas will overestimate

The evidence is strong for Mifflin-St Jeor as the default, moderate for Katch-McArdle in lean populations, and basically nonexistent for any formula being accurate within 5% for individuals without lab measurement.

The GLP-1 Problem: Why Ozempic Breaks TDEE Formulas

This is the section nobody else is writing yet, and it matters for the estimated 15 million Americans currently on GLP-1 receptor agonists.

Standard TDEE formulas assume a relatively stable ratio between lean mass and fat mass at a given weight. GLP-1 medications disrupt this assumption. Research published in Cell Reports Medicine (2025) and data from UC Davis Health show that 25-40% of weight lost during GLP-1 therapy comes from lean body mass — not just fat. In well-conducted trials with resistance training, fat mass accounted for 60-75% of total weight lost, with lean mass making up the remainder.

Here is what that means for your TDEE:

• Lean mass drives BMR. Each pound of lean tissue burns roughly 6 calories per day at rest, versus 2 calories per pound of fat.
• Disproportionate lean mass loss = faster BMR decline. If you lose 40 lbs on semaglutide and 30% is lean mass (12 lbs), your BMR drops by an extra 48 cal/day beyond what the weight loss alone would predict.
• Metabolic adaptation stacks on top. Extended caloric restriction (which GLP-1 drugs effectively create via appetite suppression) triggers adaptive thermogenesis — a 5-15% additional reduction in RMR beyond what body composition changes explain.
• Standard formulas overestimate. Mifflin-St Jeor uses total body weight without distinguishing lean vs. fat. For GLP-1 users losing lean mass at above-normal rates, the formula will overestimate TDEE by 100-300 calories per day after significant weight loss.

What GLP-1 Users Should Do

• Recalculate TDEE every 2-4 weeks during active weight loss, not every 4-6 weeks
• Use Katch-McArdle with updated body fat measurements if possible (DEXA every 3 months is ideal)
• If using Mifflin-St Jeor, subtract 5-10% from the result as a metabolic adaptation buffer
• Prioritize resistance training 3-4x/week and protein intake of 1.0-1.2g per pound of lean body mass to minimize muscle loss
• Track actual weight change against predicted change — if you are not losing at the expected rate, the formula is overestimating

Use the TDEE calculator as a starting point and the protein calculator to set your minimum protein target.

Accuracy by Sex

The 2024 comparative analysis in Medicine found that prediction accuracy varies by sex:

All formulas perform slightly better for men than women. Harris-Benedict shows the largest sex-based gap. Mifflin-St Jeor is the most consistent across sexes, which is another reason it is the recommended default.

The Activity Multiplier Problem

Even a perfectly accurate BMR formula gets undermined by the activity multiplier. The standard multipliers (1.2 for sedentary through 1.9 for extra active) were derived from limited data in the 1950s-1970s, and they introduce an additional 10-20% uncertainty.

Common mistakes with activity multipliers:

• Overestimating activity level — Three gym sessions per week does not make you “very active” if you sit 10 hours a day. Most office workers who exercise 3-5x/week are “lightly active” (1.375) or “moderately active” (1.55) at best.
• Ignoring NEAT variation — Non-Exercise Activity Thermogenesis (fidgeting, walking, posture) can vary by 500+ calories between individuals and even day-to-day within the same person. No multiplier captures this.
• Using exercise tracker calories — Wearable devices overestimate exercise burn by 20-50% on average. If you eat back “calories burned” from your watch, you are likely erasing your deficit.

The practical fix: use the calculator to get a starting estimate, then treat it as a hypothesis. Track intake and weight for 2-3 weeks. If weight is not moving as expected, adjust by 100-200 calories. The formula gets you to the right neighborhood — real-world tracking gets you to the right address.

How to Actually Use Your TDEE Number

For Fat Loss

• Set calories 300-500 below TDEE (not 1,000+ — that is how you lose muscle and crash your metabolism)
• Protein at 0.8-1.0g per pound of bodyweight — use the protein calculator
• Lift weights 3-4x/week to preserve lean mass
• Recalculate every 4-6 weeks or after every 10 lbs lost
• Expected rate: 0.5-1 lb/week fat loss

For Muscle Gain

• Set calories 200-300 above TDEE (lean bulk)
• Protein at 0.7-1.0g per pound of bodyweight
• Follow progressive overload — use the one rep max calculator to track strength
• Expected rate: 1-3 lbs/month for beginners, 0.5-1 lb/month for intermediates
• Split calories with the macro calculator

For Maintenance

• Eat at TDEE +/- 100 calories
• Track weekly weight average (not daily fluctuations)
• If weight trends up over 3+ weeks, reduce by 100 calories
• If weight trends down, increase by 100 calories

Common Misconceptions

“My metabolism is slow/fast”

Metabolic rate variation between individuals of the same weight, age, sex, and activity level is typically 200-300 calories — not the 500-1,000 calorie gap many people believe. A 2022 study in Science (Pontzer et al., n=6,421) found that after controlling for lean mass, age, and sex, individual BMR variation was remarkably narrow. Your body doesn’t read fitness magazines. It responds to energy balance, progressive stimulus, and adequate recovery.

“The more data the formula uses, the more accurate it is”

Not necessarily. The WHO/FAO equation uses only weight and age (no height) and performs comparably to Harris-Benedict, which uses all four inputs. The Katch-McArdle formula uses a single input (lean body mass) and outperforms multi-input formulas for lean individuals. The quality of the input matters more than the quantity. A precise lean body mass measurement from DEXA is more valuable than height, weight, age, and sex combined.

“I should eat exactly at my TDEE number”

Your TDEE is not a fixed number. It fluctuates by 200-400 calories day to day based on NEAT, sleep quality, stress hormones, and menstrual cycle phase (for women, BMR can vary by 100-300 cal across the cycle). Treat the number as a weekly average target, not a daily mandate.

Methodology and Sources

The primary source for accuracy comparisons is Frankenfield DC, 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;105(5):775-789. This systematic review analyzed data from studies including approximately 1,090 subjects, comparing predicted RMR to indirect calorimetry measurements.

Additional sources: Mifflin MD, St Jeor ST, et al. “A new predictive equation for resting energy expenditure in healthy individuals.” Am J Clin Nutr. 1990;51(2):241-247. Harris JA, Benedict FG. “A Biometric Study of Basal Metabolism in Man.” Carnegie Institution of Washington. 1919. Comparative analysis of basal metabolic rate measurement, Medicine (Lippincott), 2024. Pontzer H, et al. “Daily energy expenditure through the human life course.” Science. 2021;373(6556):808-812.

The GLP-1 section draws from UC Davis Health (2025) and Cell Reports Medicine (2025) research on lean mass preservation during GLP-1 therapy. These are emerging findings — the evidence base is still developing, and recommendations may change as larger longitudinal studies are completed.