How to find your maintenance calories

What maintenance actually is

Your maintenance calories are the average daily intake at which your body weight holds steady over weeks — an average against an average, not a single day. It is not a fixed constant either: it moves with body weight, activity, training, illness and sleep, which is why the useful skill is knowing how to re-measure it rather than memorising a number.

It is worth having even if you have no interest in losing or gaining weight. Maintenance is the reference point everything else is defined against: a deficit is a percentage below it, a surplus a percentage above. Without it you are adjusting from nothing.

Step 1 — get a starting estimate

Use a prediction equation. Mifflin-St Jeor, published in 1990, is the most accurate of the common ones for healthy adults; it estimates your basal metabolic rate from weight in kilograms, height in centimetres and age.

Men: (10 × weight) + (6.25 × height) − (5 × age) + 5. Women: (10 × weight) + (6.25 × height) − (5 × age) − 161.

Then multiply by an activity factor: about 1.2 sedentary, 1.375 lightly active, 1.55 moderately active, 1.725 very active. The TDEE calculator does both steps. Almost everyone overestimates their activity, so if you are between two factors take the lower one. Treat the result as a hypothesis with an error bar of a few hundred calories — prediction equations describe populations, and this one has never met you.

Step 2 — verify it over three weeks

This is the step that produces the real number, and it is the one people skip. For three weeks, do two things: eat close to your estimate on average, and weigh yourself under consistent conditions. Then compare weekly averages — average intake against average weight, never individual days. If the weight average holds flat across three weeks, your estimate was right. If it drifts, you have both a direction and a size to correct by.

Three weeks is the minimum. Week one is contaminated by whatever your body was doing before you started — a change in carbohydrate intake alone can move the scale a kilogram through glycogen and water, none of it tissue. By week three that has settled and the drift you see is signal.

• Log every day, including the days you would rather not.
• Weigh under the same conditions and compare 7-day averages.
• Do not change intake mid-experiment — that resets the clock.

A worked example

Take a 41-year-old man, 180 cm, 82 kg, desk job, gym twice a week.

BMR = (10 × 82) + (6.25 × 180) − (5 × 41) + 5 = 820 + 1,125 − 205 + 5 = 1,745 kcal. Lightly active is the honest multiplier for two gym sessions and a desk, so 1,745 × 1.375 ≈ 2,400 kcal a day. That is the hypothesis.

He eats an average of 2,400 for three weeks. His weekly weight averages come out at 82.0 kg, 82.3 kg and 82.6 kg — a drift upward of 0.6 kg, about 0.2 kg per week.

Now size the correction. The rough rule of thumb is that a kilogram of body-weight change corresponds to roughly 7,700 kcal of cumulative imbalance, so 0.2 kg per week is about 1,540 kcal per week, or about 220 kcal per day of surplus. His actual maintenance is therefore near 2,400 − 220 ≈ 2,180 — call it 2,200 and retest.

Be honest about that rule of thumb: the modelling work on energy imbalance shows the fixed-calories-per-kilogram shortcut overstates change over long horizons, because expenditure itself falls as body weight falls, and the NIH Body Weight Planner models the dynamic version properly. Over three weeks it is close enough to size an adjustment. If you would rather skip the arithmetic entirely: drifting up, cut 200 a day; drifting down, add 200; then run another three weeks.

The failure mode: trusting the calculator over the evidence

The common failure is that the scale says one thing, the calculator says another, and people believe the calculator: "my TDEE is 2,400, so if I am gaining at 2,400 my metabolism must be broken."

Almost always the explanation is simpler. Either intake is under-recorded — cooking oil, drinks, tastes and finishes, weekends logged less carefully than weekdays — or activity was overestimated at the multiplier step. Neither is a moral failure; both are measurement error, and the fix is to trust the weight data over the prediction. A useful diagnostic: if your weekday and weekend average intakes differ by more than about 300 calories, your "average" is not your average and the experiment needs a cleaner three weeks.

Using maintenance as a phase

Once you know your maintenance you can deliberately eat at it — a phase worth running, not a failure to be dieting. It keeps the tracking habit intact while removing the restriction, and it is when you re-measure, because maintenance after a period of weight change is genuinely different: a smaller body costs less to run. Spend a few weeks there, confirm the weight average has gone flat, and use that figure as the base for whatever comes next. The deficit guide covers what to do from there.

What this guide does not cover

Every number here is an estimate. Prediction equations carry meaningful error for any individual, food labels carry their own, and your logging carries more. The three-week test measures past all of that — but it measures a range, not a constant.

It also does not cover clinical cases. Thyroid conditions, PCOS, pregnancy and breastfeeding, medications affecting appetite or metabolism, eating-disorder history and adolescence all change how — and whether — this should be approached. This is general information, not medical advice; talk to a clinician or a registered dietitian if any of that applies.

FAQ