Calorie Needs: BMR, TDEE, and Why Targets Drift

A daily calorie target arrives looking like a single confident number. It is really the end of a short chain of estimates — a resting rate, an activity multiplier, a goal offset — and every link gives up a little precision. This guide walks the chain in order and uses worked numbers to show how far the answer moves when one link is set differently.

The chain, one link at a time

Three things happen between your measurements and the number you are handed. The Mifflin-St Jeor equation converts weight, height, age, and a sex constant into an estimate of energy used at rest. An activity factor multiplies that estimate to cover the rest of the day. Fixed offsets above and below the resulting maintenance figure then produce the loss and gain rows. The BMR Calculator stops after the first step, the TDEE Calculator after the second, and the Calorie Calculator runs all three — the same arithmetic exposed at three depths, not three different opinions.

Step one: the resting estimate

The equation is 10×weight(kg) + 6.25×height(cm) − 5×age + 5 for men, and the same with − 161 for women. It was published in 1990 in the American Journal of Clinical Nutrition, derived by indirect calorimetry in 498 healthy adults aged roughly 19–78 and including both normal-weight and obese participants, and it is commonly recommended over the older Harris-Benedict formula for contemporary populations.

The coefficients tell you what the equation can and cannot see. Each kilogram of body weight adds 10 kcal, each centimetre of height 6.25, each year of age subtracts 5, and the sex difference is a single flat constant rather than a separate model. There is no term for body composition and none for training history: a lean and a heavily built person of identical weight, height, and age receive identical resting estimates. That is the shape of the published formula, not a flaw in the implementation, and it is the first thing to check when an output feels wrong.

The published expectation is roughly a ±10% band around measured resting expenditure for most people, which is not a rounding error: on the example below it is about 176 kcal/day of uncertainty in the resting figure alone. Because the next step multiplies rather than adds, that uncertainty is scaled too, reaching maintenance at roughly 273 kcal/day — about the size of an entire step on the activity ladder.

Step two: the activity ladder

The engine offers five rungs, and these are the factors it applies:

• ×1.2 — Sedentary (little or no exercise)
• ×1.375 — Lightly active (1–3 days/week)
• ×1.55 — Moderately active (3–5 days/week)
• ×1.725 — Very active (6–7 days/week)
• ×1.9 — Extra active (hard exercise + physical job)

These multipliers are conventional rather than precise, and unlike the resting equation they are not measured — you choose one. Choosing between "lightly" and "moderately" active is a self-assessment that most people get wrong in the optimistic direction, and the labels compound the difficulty: they are written almost entirely in exercise days per week. Two people who both train three days a week can spend the rest of the week very differently, and the ladder has no rung for that.

The consequence is arithmetic, not opinion. The rungs are evenly spaced and the factor multiplies the whole resting estimate rather than adding to it, so the gap between two adjacent rungs is a fixed fraction of your BMR — and the larger the BMR, the larger the absolute penalty for a one-step misjudgement.

Worked example: one person, two adjacent rungs

Take a man of 80 kg and 180 cm, aged 34. Mifflin-St Jeor puts his resting estimate at 1,760 kcal/day. At the moderately active factor of 1.55, maintenance comes to 2,728 kcal/day. Move him one rung down to lightly active (1.375) — change nothing else, not a gram of weight, not a day of age — and maintenance becomes 2,420 kcal/day. The difference is 308 kcal/day, produced entirely by one dropdown.

That one choice sits between the two goal offsets the calculator applies — larger than the milder one, short of the larger — and it is about the size of the ±10% band already carried through from the resting estimate. The target is one plausible value drawn from a range at least that wide, and the activity rung is among the largest things you control about where in that range it lands.

Step three: the goal offsets

The loss and gain rows are not separately modelled. They are fixed offsets applied to maintenance: 250 kcal for the milder step and 500 kcal for the larger one, in each direction. For the moderately active example that gives 2,478 and 2,228 kcal/day for loss, 2,978 and 3,228 for gain. At the lightly active rung the same offsets land on a smaller maintenance figure, so the loss rows become 2,170 and 1,920 kcal/day.

Because the offsets are fixed rather than proportional, they mean different things to different bodies. Watch the headroom above the resting estimate: at the moderate rung the larger loss target sits 468 kcal above this person's BMR; at the lightly active rung the same offset leaves only 160 kcal. One rung of self-assessment cut that headroom to a fraction of what it was — the same fixed offset, a very different distance from the resting estimate.

Turning any of these figures into protein, carbohydrate, and fat grams is a separate convention again: the Macro Calculator applies labeled ratio splits to a calorie figure, and those splits structure intake rather than prescribe it.

The target is a hypothesis, not a prescription

The only defensible way to use the output is as the starting point of a two-to-three-week experiment against your actual weight trend. The estimate is a population formula's guess at a stranger; the trend is data about you. When the two disagree the trend is the better information, and the honest move is to re-run the chain with a rung that matches it — not with a flattering one until the calculator returns the number you wanted. It is a trend rather than a single reading because one number cannot separate a mis-set multiplier from ordinary variation, and direction and rough magnitude are exactly what the chain of estimates cannot give you.

Floors, and why the engine flags rather than hides

A loss target can fall below the person's own estimated BMR, and it can fall below the commonly cited minimum intakes — about 1,200 kcal/day for women and 1,500 for men. The engine checks both and raises a flag for each.

A concrete case: a woman of 50 kg and 155 cm, aged 60, at the sedentary factor. Her resting estimate is 1,008 kcal/day and maintenance 1,209. Subtract the standard offsets and the loss rows land at 959 and 709 kcal/day — both below the resting estimate and both below the 1,200 kcal floor. This page's build asserts that both flags are raised for exactly this input; if the engine stopped raising them, the paragraph would fail to compile rather than quietly become false.

Why flag rather than clamp? Silently raising the target to the floor would hide the fact that the arithmetic ran out of room; silently dropping the row would look like a bug. Showing the number with a warning attached is the only option that leaves the reader better informed. Targets in that territory call for qualified medical supervision rather than a calculator — a statement about what a calculator is for, not about your situation, which this page knows nothing about.

Who the equation fits least well

The study population was healthy adults, so the equation is least reliable for teenagers, athletes at body-composition extremes, and people with conditions that alter metabolism. The engine encodes part of that boundary directly: it accepts ages of 15–100 for BMR and calories, weights of 10–500 kg, and heights of 50–280 cm, returning an explicit message rather than a wrong answer outside them. The age floor is a hard edge, not a warning.

For the body-composition case there is no fix inside this chain, because the equation has no composition term to correct. A circumference estimate from the Body Fat Calculator at least gives a second, independent number to hold alongside it; what a height-and-weight ratio can and cannot see is covered in what BMI does and does not measure.

Three common misreadings

Adding workout calories on top

The activity multiplier is already meant to cover training — that is what it is for. Adding a watch's estimate of a session's burn to a TDEE figure counts the same exercise twice. To work from a base and add sessions separately, start from the resting estimate, not from maintenance.

Reading the whole number as precision

Calories are displayed as whole numbers because that is the precision the estimates deserve; the arithmetic keeps full precision internally and rounds only at display. Against the roughly 273 kcal/day that the resting band alone contributes at the example's maintenance figure, a handful of kcal between two targets carries no information.

Assuming the number keeps

Weight is an input, so the output moves with it. Each kilogram carries a coefficient of 10 kcal in the resting equation, which the moderate multiplier scales to 15.5 kcal in the maintenance figure. The target has a date on it, and re-running the chain after a real change in weight is part of using it, not a sign that the first answer was wrong.

Frequently asked questions

Not medical advice: this guide provides general educational information about published population formulas. It cannot account for your individual health context and must not be used to diagnose or treat any condition — consult a qualified healthcare professional before changing your diet, especially with any medical condition, pregnancy, or a history of disordered eating. Calculations are processed locally in your browser and never transmitted. See the methodology page.