Behind every TDEE estimate sits a metabolic equation — a mathematical model built from laboratory measurements of real people breathing into calorimeters. Three formulas dominate the field: Mifflin-St Jeor (1990), the revised Harris-Benedict (1984), and Katch-McArdle (1983). They were developed decades apart, derived from different populations, and built on different assumptions about what drives resting metabolism. Most online calculators pick one and never mention the others. That obscures a meaningful question: how much do they actually diverge, and does it matter for your nutrition planning?
The short answer is that formula choice typically shifts BMR estimates by 50 to 200 kcal per day — a range that compounds across weeks but is often smaller than the uncertainty introduced by activity level selection. The longer answer requires understanding where each equation came from, what it was designed to predict, and where the evidence points today. That context turns a black-box number into something you can actually evaluate.
A Brief History of Metabolic Prediction
Metabolic prediction equations did not appear in a vacuum. Each emerged from a specific research programme, used the measurement technology available at the time, and reflected the population the researchers had access to. Tracing that lineage explains why the formulas disagree — and why none of them can claim universal accuracy.
In 1919, James Arthur Harris and Francis Gano Benedict published what would become the most cited metabolic equation in history. Working at the Carnegie Nutrition Laboratory in Boston, they used indirect calorimetry — measuring oxygen consumption and carbon dioxide production — on 239 subjects. The resulting equations expressed BMR as a function of height, weight, age, and sex. For decades, the Harris-Benedict equations were the only game in town, embedded in clinical practice and dietetic training worldwide.
By the 1980s, researchers had accumulated enough validation data to identify systematic problems. The original Harris-Benedict equations tended to overestimate BMR, particularly in certain subgroups. In 1984, Roza and Shizgal published a revised version using updated statistical regression techniques and a more diverse subject pool. The revised coefficients reduced prediction error and became the version most commonly labelled "Harris-Benedict" in modern calculators — though many sources fail to specify whether they mean the 1919 original or the 1984 revision.
A year earlier, Frank Katch and William McArdle had taken a fundamentally different approach. Rather than predicting BMR from total body weight (which bundles metabolically active muscle with relatively inert adipose tissue), they built an equation around LBM — the portion of body weight that is not fat. The Katch-McArdle formula is elegantly simple: 370 + 21.6 multiplied by lean mass in kilograms. Its strength is theoretical precision; its weakness is that it requires a body fat percentage measurement, adding another layer of estimation before you even reach BMR.
Then in 1990, Mifflin and St Jeor published a new set of equations in the American Journal of Clinical Nutrition, derived from 498 subjects spanning a wider age and weight range than earlier studies. Their equations retained the height-weight-age-sex structure of Harris-Benedict but with updated coefficients that proved more accurate in subsequent validation work. A landmark 2005 systematic review by Frankenfield and colleagues, published in the Journal of the American Dietetic Association, found that Mifflin-St Jeor predicted resting energy expenditure within 10% of measured values for more subjects than any competing equation.
That historical arc — from Harris-Benedict's pioneering 1919 work through the lean-mass innovation of Katch-McArdle to Mifflin-St Jeor's larger, more representative dataset — shapes how the formulas perform today. They are not interchangeable. Each carries the fingerprint of the population and methodology that produced it.
The Three Formulas Side by Side
Seeing the actual equations clarifies what each formula uses as inputs and how the mathematics differ. All three produce a BMR or RMR estimate in kilocalories per day. To convert BMR to TDEE, each result is then multiplied by an activity factor — a step covered in detail below.
Mifflin-St Jeor (1990):
Males: (10 × weight in kg) + (6.25 × height in cm) − (5 × age in years) + 5
Females: (10 × weight in kg) + (6.25 × height in cm) − (5 × age in years) − 161
Harris-Benedict Revised (Roza & Shizgal, 1984):
Males: 88.362 + (13.397 × weight in kg) + (4.799 × height in cm) − (5.677 × age in years)
Females: 447.593 + (9.247 × weight in kg) + (3.098 × height in cm) − (4.330 × age in years)
Katch-McArdle (1983):
370 + (21.6 × lean body mass in kg)
The structural difference is immediately visible. Mifflin-St Jeor and Harris-Benedict both use total body weight, height, age, and sex as predictors. Katch-McArdle collapses all of that into a single variable — lean body mass — making it sex-agnostic and age-agnostic (the assumption being that LBM already captures those effects). This economy is both its appeal and its limitation: the formula is only as accurate as the body fat estimate feeding it.
To make the comparison concrete, consider a 30-year-old male weighing 80 kg at 178 cm tall with an estimated body fat of 18%. His lean body mass would be 65.6 kg (80 × 0.82). The following table shows each formula's BMR estimate and the resulting TDEE at two common activity levels.
| Formula | BMR (kcal/day) | TDEE Sedentary (×1.2) | TDEE Moderate (×1.55) |
|---|---|---|---|
| Mifflin-St Jeor | 1,780 | 2,136 | 2,759 |
| Harris-Benedict (revised) | 1,842 | 2,210 | 2,855 |
| Katch-McArdle | 1,787 | 2,144 | 2,770 |
For this individual, all three BMR estimates fall within a 62 kcal range (1,780 to 1,842 kcal). At the sedentary activity level, TDEE estimates span about 74 kcal; at moderate activity, roughly 96 kcal. Those gaps are real, but they are modest compared to the 600+ kcal swing created by changing the activity multiplier from sedentary to moderate. You can run your own numbers through the TDEE Calculator with three-formula comparison to see how your profile shifts the results.
What Validation Studies Tell Us
An equation's pedigree matters less than its performance in independent validation. Fortunately, all three formulas have been tested against measured metabolic rates across diverse populations, and the evidence paints a reasonably clear picture.
The most influential review is Frankenfield, Roth-Yousey, and Compher's 2005 systematic analysis published in the Journal of the American Dietetic Association. After evaluating 173 studies, they concluded that the Mifflin-St Jeor equation predicted RMR within 10% of measured values in the highest proportion of non-obese and obese individuals. On the strength of this finding, the American Dietetic Association (now the Academy of Nutrition and Dietetics) recommended Mifflin-St Jeor as the preferred equation for estimating RMR in healthy adults.
Harris-Benedict, particularly the original 1919 version, was found to overestimate RMR by 5–15% in several populations. The 1984 revision reduced but did not eliminate this bias. Clinical settings continue to use Harris-Benedict frequently — in part because of institutional inertia and in part because clinicians are familiar with its behaviour and can apply judgment-based corrections. For most non-clinical purposes, though, Mifflin-St Jeor is the stronger starting point.
Katch-McArdle occupies a different niche. Because it uses lean body mass, its accuracy is tightly coupled to the quality of the body composition measurement. When body fat is measured via DEXA or hydrostatic weighing, Katch-McArdle performs well — particularly at the extremes of body composition where weight-based formulas struggle most. When body fat is estimated from skinfold calipers or bioelectrical impedance (both of which carry their own error margins), the compounding uncertainty can make Katch-McArdle less reliable than Mifflin-St Jeor despite its theoretical advantage.
A practical takeaway from the validation literature is that no single formula wins in every scenario. Mifflin-St Jeor has the broadest applicability, Harris-Benedict has the longest track record (with known limitations), and Katch-McArdle offers a different analytical lens when lean mass data is available. Using the dedicated BMR comparison tool to view all three results simultaneously provides more context than any single number could.
When Each Formula Shines
Knowing the validation data is useful, but translating it into a decision — which formula should you actually pay attention to? — requires matching each equation's strengths to specific situations.
Mifflin-St Jeor: The General-Purpose Standard
For most people in most circumstances, Mifflin-St Jeor is the appropriate default. It does not require a body fat measurement, it has been validated across a wide weight range, and it was derived from a reasonably diverse sample. If you have only height, weight, age, and sex to work with, Mifflin-St Jeor is the equation with the strongest evidence base behind it.
Its main limitation is the same one shared by Harris-Benedict: it uses total body weight. Two people who weigh 90 kg — one carrying 15% body fat and the other carrying 30% — will receive identical BMR estimates from Mifflin-St Jeor despite having very different metabolic profiles. For the general population, this limitation is usually within acceptable error bounds. For athletes and individuals at the extremes of body composition, it may introduce meaningful inaccuracy.
Harris-Benedict: The Historical Benchmark
The revised Harris-Benedict equation remains widely used, particularly in healthcare settings where it is embedded in electronic health records and institutional protocols. Its tendency to produce slightly higher estimates than Mifflin-St Jeor is well documented, and experienced practitioners often account for this when interpreting results.
Harris-Benedict also serves a useful role as a comparison point. If Mifflin-St Jeor and Harris-Benedict agree closely for a given individual, confidence in the estimate is higher. If they diverge significantly, it may signal that the individual falls outside the population characteristics either formula was designed around — a prompt to gather more data or seek professional assessment.
Katch-McArdle: The Lean Mass Specialist
Katch-McArdle is the only formula in the standard trio that accounts for body composition directly. For individuals who have a reliable body fat measurement — from DEXA, hydrostatic weighing, or a well-performed skinfold assessment — Katch-McArdle can provide a more nuanced estimate than either weight-based equation.
This advantage is most pronounced at the extremes. A competitive bodybuilder at 5% body fat and a sedentary individual at 35% body fat may have similar total body weights but radically different metabolic rates. Mifflin-St Jeor and Harris-Benedict will treat them similarly; Katch-McArdle will not. If you have body composition data, use the body fat estimation for Katch-McArdle input to ensure the number feeding the formula is as accurate as possible. For guidance on measurement technique, consult the guide to measuring body fat accurately.
The Activity Multiplier Problem
Most discussions of TDEE accuracy focus on the BMR formula — but the activity multiplier applied to BMR introduces far more uncertainty than the choice between Mifflin-St Jeor, Harris-Benedict, and Katch-McArdle combined.
The standard activity multipliers, originally adapted from work by Harris and Benedict and later codified in various nutritional guidelines, are broad categories.
| Activity Level | Multiplier | Description |
|---|---|---|
| Sedentary | 1.2 | Desk job, minimal structured exercise |
| Lightly active | 1.375 | Light exercise 1–3 days per week |
| Moderately active | 1.55 | Moderate exercise 3–5 days per week |
| Very active | 1.725 | Hard exercise 6–7 days per week |
| Extremely active | 1.9 | Physical job plus daily training |
For someone with a BMR of 1,700 kcal, the difference between sedentary (2,040 kcal TDEE) and very active (2,933 kcal TDEE) is 893 kcal per day. By contrast, the three BMR formulas for the same individual typically fall within a 50–200 kcal range. The multiplier gap is four to fifteen times larger than the formula gap.
This mismatch in error magnitudes has a practical implication: agonising over which BMR formula to use while guessing at your activity level is optimising the wrong variable. Most people overestimate their activity level by at least one tier. Someone who trains hard three times per week but spends the remaining hours at a desk is closer to "lightly active" than "moderately active," because NEAT — the energy burned through fidgeting, walking, and all non-exercise movement — often contributes more to TDEE than structured exercise sessions.
The most effective approach is to select an activity level conservatively, calculate a TDEE range using multiple formulas, and then calibrate against real-world weight trends over two to four weeks. A predicted TDEE that does not match actual weight changes is not wrong in a mathematical sense — it simply needs adjustment based on individual data that no population-level formula can capture.
Practical Recommendations
Distilling the research and the formula mechanics into a decision framework produces a clear hierarchy, along with a protocol for improving accuracy over time.
Start with Mifflin-St Jeor. It has the broadest validation, does not require body fat data, and was specifically recommended by the American Dietetic Association for general use. If you have no body composition measurements and need a single number to work from, Mifflin-St Jeor is the strongest starting point.
Add Katch-McArdle if body fat data is available. When you have a body fat estimate from a reasonably reliable method (DEXA, hydrostatic weighing, a well-calibrated skinfold protocol, or even the Navy tape method), Katch-McArdle provides a useful second data point. If it agrees closely with Mifflin-St Jeor, your confidence in the estimate increases. If the two diverge significantly, investigate the body fat input — a measurement error there will cascade through the Katch-McArdle result.
Run all three for comparison. The TDEE Calculator with three-formula comparison displays Mifflin-St Jeor, Harris-Benedict, and Katch-McArdle side by side, which immediately reveals how much (or how little) formula choice matters for your specific profile. Treat the range between the lowest and highest as a confidence interval rather than picking one number.
Once you have a TDEE estimate, the next step is translating it into nutrition targets. Use the macro calculator to convert TDEE into actionable nutrition targets for daily protein, carbohydrate, and fat breakdowns. If your goal involves weight change, the deficit planning tool that builds on your TDEE estimate can set an appropriate daily calorie target. For those focusing on recovery and muscle protein synthesis, understanding protein requirements tied to your overall energy balance adds another layer of precision.
Most importantly, treat any TDEE estimate as a hypothesis to be tested, not a verdict. Weigh yourself under consistent conditions (morning, fasted, after using the bathroom), track the seven-day moving average, and compare against predicted changes over two to four weeks. If your weight is stable and your TDEE estimate says you should be losing, the estimate is too high — adjust downward by 100–200 kcal and re-evaluate. This iterative calibration will always outperform any formula alone.
The Bottom Line
No metabolic formula is perfect. Harris-Benedict, Katch-McArdle, and Mifflin-St Jeor are all population-level regression models applied to individuals — and individuals are messy. Genetics, hormonal status, medication use, sleep quality, daily movement patterns, and even ambient temperature all influence real metabolic rate in ways that height, weight, and age cannot capture.
The practical value of comparing three formulas is not in finding the "right" one. It is in understanding the range of plausible estimates, recognising that the activity multiplier introduces more uncertainty than the BMR formula, and building a feedback loop between calculated estimates and real-world outcomes. A TDEE estimate that is refined through two to four weeks of tracking data will always be more accurate than any formula used in isolation — regardless of which equation produced the initial number.
The formulas matter, but they are the starting point, not the destination.