Maintenance is a number your body reports, not one a formula predicts#
Your maintenance calories are the intake that holds your weight steady over time, and the only dependable way to find them is to measure, not to predict. Track what you eat, keep it roughly constant for two to four weeks, watch the weekly weight trend, and back out the answer: if the scale held, that intake is your maintenance; if it drifted, the size of the drift tells you by how much you missed and in which direction. A calculator gives you a place to start. Your own data gives you the number.
The reason the measurement beats the prediction is not that calculators are badly built — it's that they are built on other people. Even the best-validated resting-metabolism equations leave about one adult in five outside a plus-or-minus 10 percent band before any activity multiplier is applied, as how to calculate your TDEE lays out, and two people of identical height, weight, age and sex can still burn meaningfully different amounts each day (TDEE explained has why). An equation cannot see your job, your commute, your fidgeting or your dieting history. Your weight trend integrates every one of them, automatically, without knowing their names. This article is that measurement method: the arithmetic, how long to run it, and the two traps that make people misread the result.
What "maintenance" actually means on a scale#
Strip away the calculators and maintenance has a plain physical definition. Over any window of time, the change in the energy your body has stored equals what you ate minus what you burned. Store energy and the scale rises; spend it and the scale falls. Maintenance is simply the intake at which those two cancel across a long enough window that day-to-day noise averages out.
That definition is also the method. You don't need to know your burn to find the intake that matches it — you need to feed your body a known, steady amount and let it tell you which side of the line that amount sits on. The scale is the readout. You are running a two-to-four-week experiment with a sample size of one, and the one is the only one that counts.
The back-calculation, step by step#
Three steps, and the third is arithmetic you can do on a phone.
Step 1 — eat a consistent, tracked amount for two to four weeks. Pick a plausible starting intake (a calculator's number is fine as a hypothesis) and hold it steady. Log the same way every day — same app, same habits, same weighing-or-guessing of portions. Consistency matters far more than accuracy here, for a reason that becomes the whole point in a moment. How to count calories covers doing it repeatably.
Step 2 — track weight as a weekly average, under the same conditions. Weigh most mornings, after the bathroom, before eating or drinking, and average each week. Single days are almost pure noise; the weekly average is the signal.
Step 3 — back out the number. Convert your weight trend into a daily energy imbalance and adjust your intake by it. A kilogram of body-weight change is worth roughly 7,700 kcal (about 3,500 kcal per pound) — the energy density behind the calories in a pound of fat. So:
Maintenance ≈ your average daily intake + (weight change in kg × 7,700 ÷ number of days). Lose weight and you add the difference back; gain weight and you subtract it.
The arithmetic below is ours, worked on that 7,700 kcal/kg figure — not a value any single study reports, but a well-founded conversion:
| Over the window you… | Ate (daily) | Weight change | Daily imbalance (our arithmetic) | Estimated maintenance |
|---|---|---|---|---|
| Held steady (21 days) | 2,200 | 0.0 kg | 0 | ~2,200 |
| Lost weight (21 days) | 2,000 | −0.5 kg | −183 | ~2,180 |
| Gained weight (28 days) | 2,600 | +0.4 kg | +110 | ~2,490 |
One caution on that conversion, and it is the reason not to extrapolate it: the 7,700 figure describes the tissue, not a promise about the future. The dynamic-modelling work is explicit that expenditure falls as you lose, so the old "cut 500 a day, lose a pound a week forever" rule overstates real long-run loss by roughly 100 percent, with body weight taking about a year to fully settle after any permanent change in intake3. Back-calculation sidesteps that trap precisely because it measures the imbalance you are running right now rather than predicting a trajectory — but it also means the number has a shelf life, which the last section returns to.
Why two to four weeks, and never one#
The single biggest mistake is reading the number too soon. Body weight over a few days is mostly water, and water moves for reasons that have nothing to do with fat.
Most of that movement rides on glycogen, your stored carbohydrate. Glycogen is warehoused in hydrated form — three to four grams of water bound to every gram of glycogen4. Change your carbohydrate intake, your salt, your training, or simply the day, and you can swing a kilogram or two on the scale in 24 hours without having gained or lost a meaningful amount of tissue. That is why a single week can lie in either direction: it can show a "loss" that is a drained glycogen store, or a "gain" that is a salty dinner and a full gut.
A multi-week window fixes this because the water shift is largely a one-time step. When you first hold a new intake steady, glycogen and its water find a new level within the first several days and then stop moving — so the slope of weeks two through four reflects tissue, not plumbing. The practical rule: ignore the first few days after any change, weigh through at least two full weeks, and trust the weekly averages, not the daily readouts. If your schedule swings hard between weekdays and weekends, run the full four.
The underreporting trap — and why it doesn't sink the method#
Here is the objection that should worry you, because it's true: people are bad at logging food, and the errors run one way. When researchers measured 224 obese adults with doubly labeled water, the subgroup who "couldn't lose weight on 1,200 calories" were in fact eating an average of 2,081 — underreporting their intake by 47 ± 16 percent, and over-reporting their exercise by 51 ± 75 percent, with faulty memory rather than dishonesty as the mechanism1. The gap is not a quirk of that one group, either: across the validation literature, no self-report instrument is accurate against doubly labeled water, and systematic under-reporting is the rule2.
So if your log reads low, doesn't your maintenance estimate read low too? It does — and it doesn't matter, because of what the number is for. Back-calculation produces a maintenance figure denominated in your own logging, not in true physiological calories. If your habits undercount by a steady fraction, the maintenance number you derive is undercounted by the same fraction — and you will then eat to it using that same habit, so the two errors land on top of each other and cancel. The figure is personal, non-transferable, and only valid while you keep logging the way you did when you measured it. Change apps or start weighing your oats and you have changed the units; re-run the loop.
That is the deeper reason consistency beats accuracy. A precise logger and a habitual under-counter, tracking the same body, will arrive at different maintenance numbers and both will be right for themselves. What you are calibrating is not the truth about your metabolism — it's the dial on your own instrument.
Your maintenance is a moving target#
One last thing to own: the number you find is a snapshot, not a constant. It drifts down as you lose weight — partly because a smaller body costs less to run, and partly through the modest, real metabolic adaptation covered in is 'starvation mode' real. And because weight takes months to fully equilibrate to an intake change3, the maintenance you measure in January is a fact about January.
So treat the loop as repeatable, not one-and-done. Re-derive your maintenance every 4-5 kg of change, or every couple of months during an active calorie deficit, and expect the figure to fall as you go rather than reading its decline as failure. The equation you started with never gets more accurate. Your own measurement does, every time you run it.
FAQ#
How long does it take to find your maintenance calories?#
Two to four weeks of consistent tracking is the realistic minimum. Shorter than that and water and glycogen shifts — three to four grams of water swing with every gram of glycogen — swamp the real trend, so a single week can show a phantom loss or gain of a kilogram or more. Weigh most mornings under the same conditions, use the weekly average rather than daily readings, and ignore the first few days after any change in intake while the water settles.
What if I lose weight while eating at my calculated maintenance?#
Then the calculator's number was above your true maintenance, and your own data just corrected it — this is the method working, not failing. Convert the loss: about 7,700 kcal per kilogram, divided by the number of days, is your daily deficit. Add that figure to what you were eating and you have a better maintenance estimate. Hold the new number for another two weeks and confirm the scale is flat.
Why is my real maintenance so different from the online calculator's?#
Because the calculator is fitted to a population and you are one person. The best resting-metabolism equations miss by more than 10 percent for roughly one adult in five, and that error is then multiplied by an activity category you chose off a dropdown rather than measured. Your job, your daily movement and your dieting history all move your real maintenance and are invisible to the formula — which is exactly why a measured trend beats a predicted number.
Sources#
- Lichtman SW, Pisarska K, Berman ER, et al. Discrepancy between self-reported and actual caloric intake and exercise in obese subjects. N Engl J Med. 1992;327(27):1893-1898.
- Trabulsi J, Schoeller DA. Evaluation of dietary assessment instruments against doubly labeled water, a biomarker of habitual energy intake. Am J Physiol Endocrinol Metab. 2001;281(5):E891-E899.
- Hall KD, Sacks G, Chandramohan D, et al. Quantification of the effect of energy imbalance on bodyweight. Lancet. 2011;378(9793):826-837.
- Kreitzman SN, Coxon AY, Szaz KF. Glycogen storage: illusions of easy weight loss, excessive weight regain, and distortions in estimates of body composition. Am J Clin Nutr. 1992;56(1 Suppl):292S-293S.
