Why Two Calorie Calculators Give You Different Numbers

The equations barely disagree. The dropdown you fill in yourself is where the error lives.

By SimpleTools Daily editorial team · Published 2026-09-10 · Updated 2026-09-10 · How we check this

Put the same person into three calorie calculators and you will get three different numbers. Take a 40-year-old man, 82 kg, 178 cm, who tells each of them he is moderately active. The first returns about 2,694 calories a day. The second returns 2,812. The third, which additionally asks for his body fat percentage, returns 2,770. Nobody has made an arithmetic mistake; they are using three different equations for the same quantity. The interesting thing is not that they disagree, but by how little—and how much larger the error is in the one input he supplied himself.

Three equations, one body

Nearly every TDEE calculator works in two steps. First it estimates basal metabolic rate (BMR)—roughly, the energy cost of being alive and at rest. Then it multiplies that by an activity factor to reach total daily energy expenditure. The equations in common use for step one are these three:

• Harris-Benedict, originally published in 1919 and revised by Roza and Shizgal in 1984. It uses weight, height, age and sex.
• Mifflin-St Jeor, published in 1990 in the American Journal of Clinical Nutrition, fitted to a more contemporary population. Same four inputs, different coefficients.
• Katch-McArdle, which ignores sex, height and age entirely and uses only lean body mass: BMR = 370 + 21.6 × LBM(kg).

Run our 82 kg, 178 cm, 40-year-old man through all of them:

The spread across the four rows is about 116 calories, or 6.8% of the smallest value (1,698). That is the honest size of the "which equation" question for a typical adult. It is not zero, and there is a defensible ranking: a 2005 systematic review by Frankenfield, Roth-Yousey and Compher in the Journal of the American Dietetic Association concluded that Mifflin-St Jeor "was the most reliable, predicting RMR within 10% of measured in more nonobese and obese individuals than any other equation, and it also had the narrowest error range." That is why the TDEE Calculator here uses it. But note the framing even in the favourable result: within 10% of measured, for most people. The equation that wins is still routinely off by a tenth in either direction for an individual, because it is a regression line through a scatter of real humans and you are one point in that scatter, not the line.

The Katch-McArdle trap

Katch-McArdle is often presented as the most accurate of the three, on the reasonable theory that lean tissue is what burns the calories, so an equation keyed to lean mass should beat one keyed to total weight. Physiologically that is sound. Practically it frequently makes the estimate worse, and the table above shows why: moving the assumed body fat percentage from 20% to 25% changes the output by 89 calories. That single assumption moves the answer nearly as much as the entire gap between Mifflin-St Jeor and Harris-Benedict.

Unless you have a DEXA scan, your body fat figure came from a caliper reading, a bathroom scale sending a current through your feet, or a tape-measure formula—all of which carry error bars of several percentage points. An equation with a better model and a worse input is not more accurate. It is more precisely wrong. This is a general property of estimation that gets ignored constantly in fitness contexts: the quality of a formula is capped by the quality of the number you feed it.

The activity multiplier dwarfs everything else

Here is the part that makes the equation debate close to irrelevant. Take the Mifflin-St Jeor BMR of 1,738 and apply the standard multiplier ladder:

One step on that ladder—from "lightly active" to "moderately active"—is worth about 300 calories a day. That is two and a half times the entire spread between all three BMR equations. Across the full range, 1.2 to 1.9, the multiplier changes the answer by 58%.

And this enormous term is the one input nobody measures. You pick it from a dropdown with labels like "exercise 3-5 days per week" that require you to average your own behaviour over a period you are not tracking, in categories that overlap. Someone who trains hard four times a week but sits at a desk the rest of the time and someone who never trains but works on their feet in a warehouse can plausibly select the same band and have genuinely different expenditure. The bands themselves descend from the physical activity level (PAL) concept used in the energy-requirement literature—the National Academies' 2023 Dietary Reference Intakes for Energy uses PAL categories in the same spirit—but a PAL derived from doubly labelled water measurement and a PAL chosen from a dropdown are not the same quantity, even when they carry the same number.

NEAT: the variance no equation contains

There is a further source of spread that none of these models even attempt to represent. In 1999, James Levine and colleagues published a study in Science in which 16 nonobese volunteers were overfed 1,000 calories a day above maintenance for eight weeks. Fat gain differed roughly tenfold across the group. The largest single explanation was non-exercise activity thermogenesis—fidgeting, posture, spontaneous movement, the unaccounted motion of ordinary life—which accounted for about two-thirds of the increase in total daily energy expenditure and varied enormously between individuals.

NEAT does not appear in Mifflin-St Jeor, Harris-Benedict or Katch-McArdle, and it is not what the activity multiplier is asking about, since the dropdown is worded in terms of deliberate exercise. Two people with identical height, weight, age, sex, body composition and gym schedule can differ by several hundred calories a day through this channel alone. This is contested territory in the details—researchers disagree about how much of the variation is stable individual trait versus adaptive response to energy load, and about how well short-term overfeeding studies generalise—but the existence of large between-person variation that the equations cannot see is not seriously in dispute.

Use the number as a hypothesis, not a target

All of which points to a specific way of using a TDEE estimate that is different from how most people use it. The output is not your calorie requirement. It is a starting guess with an uncertainty band of a few hundred calories, and its only real job is to give you somewhere to begin measuring from.

The practical protocol:

1. Take the estimate. Eat at approximately that intake, consistently, for two to three weeks.
2. Weigh yourself under the same conditions daily and use the weekly average, not individual readings. Day-to-day weight moves several pounds on glycogen, sodium, digestive contents and menstrual-cycle fluid shifts, all of which swamp a real energy imbalance over short windows. That noise is exactly why the observation period needs to be two or three weeks rather than three days.
3. Compare the trend to your intention. As a rough working figure, an average change of about 0.8 kg over three weeks corresponds to an energy imbalance in the region of 300 calories a day—which is to say, roughly one wrong step on the activity ladder.
4. Adjust the intake, not the calculator. Then repeat.

Treat the popular "3,500 calories per pound" rule with suspicion when projecting forward, incidentally. It describes the energy content of stored fat reasonably well but assumes expenditure stays constant as you lose weight, which it does not—a lighter body costs less to move and maintain, so a fixed deficit produces a decelerating curve rather than a straight line. The NIH's Body Weight Planner exists largely because linear extrapolation overpredicts long-run loss.

None of this is medical advice, and it is not a plan for anyone managing a condition, taking medication that affects appetite or metabolism, recovering from an eating disorder, pregnant, or working with a clinician—in those situations the number from a web form is not the input that should be driving decisions. For everyone else, the useful shift is this: stop looking for the calculator that gives the right number. The scale gives the right number. The calculator just tells you where to start looking.

If you want the supporting estimates, TDEE gives you the starting hypothesis, BMI is worth reading with its own well-known limitations in mind, and water intake is a reminder that hydration state is a large part of why the scale is noisy in the first place.

Related tools

• TDEE Calculator — Mifflin-St Jeor BMR with activity multipliers
• BMI Calculator — a ratio, not a verdict
• Water Intake Calculator — daily hydration estimate

Related tools

• TDEE Calculator
• BMI Calculator
• Water Intake Calculator