Your Maintenance Calories Are Probably Wrong By 200 a Day, and You Can Prove It in Three Weeks

Every energy expenditure formula you have used is a regression equation fitted to a population and then handed to you as if it described you specifically.

That is not a criticism of the formulas. It is a description of what they are, and it has a consequence worth knowing: the standard error on those equations is large enough that your starting target is probably wrong by a couple of hundred calories a day in one direction or the other, and you have no way of knowing which.

The good news is that the fix requires no new information. You are already collecting it.

The arithmetic

Solving for maintenance from your own data

Average daily intake over 3 weeks: I

Weight change over the same window: ΔW (kg, from weekly averages)

Energy per kg of body mass change: ≈ 7,700 kcal

Maintenance M = I − (ΔW × 7,700 / 21)

Worked: intake 2,600 kcal, down 0.6 kg over 21 days → M = 2,600 − (0.6 × 7,700 / 21) = 2,600 − 220 = 2,380 kcal.

If your formula told you 2,600 and this says 2,380, you have been eating at maintenance while believing you were in a deficit. Nothing was broken. The estimate was.

The two things that break this

Daily weigh-ins. Body weight moves several hundred grams to a kilogram day to day on water, glycogen and gut contents alone, and that noise is larger than the fat mass you can gain or lose in a day. Use a weekly average. Two weeks is the earliest a trend is legible; three is where it becomes reasonably reliable.

Sloppy intake logging. The arithmetic above is only as good as I, and this is where most attempts fail. If your intake number carries a 12% error — which is where the loosest mainstream trackers sit — then on 2,600 kcal you have ±310 kcal of uncertainty in the input, which is larger than the correction you are trying to detect.

That is the practical reason to care which tracker you use, and it is a stronger argument than any spec sheet: the accuracy of your food log determines whether this calculation can work at all.

What “7,700 kcal per kg” hides

It is an approximation for mixed tissue, and it is fine for this purpose over three weeks. It is less fine over three days, where the composition of the change is mostly not fat, and less fine over a very aggressive deficit, where lean mass loss shifts the ratio.

Use it as what it is: a workable constant for a three-week window, not a law.

Why we bother writing this down

Because the alternative is what most people do, which is to pick a formula, follow it, get an unexpected result, and conclude something is wrong with their metabolism.

Almost always the metabolism is behaving exactly as arithmetic predicts, and the error is in an estimate somebody handed you at setup. Three weeks of your own data replaces it.

Questions we get asked

How do I calculate my maintenance calories accurately?

Not from a formula. Log your intake for three weeks, take a weekly average of your body weight rather than daily readings, and solve backwards: maintenance equals your average intake minus the change in stored energy implied by your weight trend, using roughly 7,700 kcal per kilogram of body mass. A formula gives you a starting guess with a standard error of a few hundred calories; three weeks of your own data gives you your number.

Why do calorie calculators disagree with each other?

Because they are different regression equations fitted to different populations. Mifflin-St Jeor, Harris-Benedict and Katch-McArdle will all give you slightly different answers for the same body, and none of them is wrong — they are estimates of a population mean, and you are one person with your own non-average activity, NEAT and lean mass. The disagreement between calculators is a good illustration of the size of the uncertainty.

How long before I should change my calorie target?

Three weeks minimum, and use weekly averages. Daily weight noise is dominated by water, glycogen and gut contents, and it comfortably exceeds the fat mass you can gain or lose in a day. Two weeks is the earliest a trend is legible; three is where it is reasonably reliable. Changing your target every few days means you are chasing measurement noise.

Tessa Lindqvist

Contributing writer, nutrition · Stronger by Math

Writes about intake, adherence and what the self-monitoring literature actually supports. Weighs her own food, which is why she is sceptical of estimates — including the good ones.