How to Find Your Real Maintenance Calories
Online calculators guess your maintenance calories from formulas built on population averages and a self-reported activity guess. The measured alternative: log honestly, weigh daily, and average by week until two weeks land within a pound of each other.
Your real maintenance calories are not a number a calculator can give you. They are whatever you were actually eating during the stretch of weeks your weight held flat, discovered after the fact rather than predicted ahead of time. A formula estimates your resting calories from your height, weight, age and sex, multiplies that by a guess about how active you are, and hands you a number that was never checked against how your body actually behaves. The number your own log produces, measured rather than modeled, is usually the one that turns out to be right.
Why a calculator's number is a guess, not a measurement
Every online calculator does roughly the same three things. It estimates your resting calories with an equation built from population averages, multiplies that by a number standing in for how active you are, and calls the result your maintenance calories. The equation itself is doing reasonable work. A 2005 review that checked several resting-calorie equations against directly measured values in hundreds of people found the best-performing one, the Mifflin-St Jeor equation, was the most consistently accurate of those tested (Frankenfield, Roth-Yousey & Compher, 2005).
"Most accurate of the group" is not the same claim as "correct for you". Two people can share a height, weight, age and sex and still burn a genuinely different number of resting calories, because muscle mass, thyroid output, and the energy spent on fidgeting and posture across a day are not identical between them, and none of those variables appears anywhere in the equation. Swap in a different formula, Harris-Benedict, Katch-McArdle, whichever one a given site happens to have coded, and the estimate moves, sometimes by a meaningful amount, without your actual metabolism changing at all. That spread between formulas is itself a clue that none of them is measuring you specifically.
The activity multiplier does more damage than the resting-calorie equation ever could. "Lightly active" and "moderately active" are self-reported categories with no shared definition behind them. Someone who trains three times a week and sits at a desk the other sixteen hours is a different case from someone who trains three times a week and is on their feet for an eight-hour shift, and a dropdown menu cannot tell those two people apart. It picks a multiplier and hopes.
What does "maintenance calories" actually mean?
Maintenance calories are not a fixed property of your body the way your height is. They describe a relationship between two numbers over a stretch of time: what you ate, and what your weight did in response. If your weight was the same at the end of three weeks as it was at the start, whatever you ate across those three weeks was, by definition, close to what your body used across that same stretch. No formula, no lab and no metabolic test is required to know that. The flat line on the scale is the proof.
This is also why the number does not stay put. Add real muscle over a training block and maintenance calories rise a little, because muscle tissue costs more to maintain than the fat it displaced. Sleep badly for a month and the number can drop. A figure measured in March is a fact about March. It is not a fact about you that holds indefinitely, and treating it as one is the same mistake as trusting the calculator, just with better evidence behind the wrong belief.
The measured method: log, weigh, and average by week
Log what you actually eat, as accurately as you can manage, in your food log for two to four weeks. Weigh yourself most mornings under matching conditions: same rough time, same amount of clothing, ideally before food or water. Average the mornings into weekly numbers rather than reading any single day on its own, because a single morning's weight is mostly noise. A pound or two of overnight swing is normally water, not fat, driven by glycogen refilling, a salty meal still working its way through, or, for many people, a shift tied to the menstrual cycle running on an eleven-to-fourteen-day rhythm of its own.
None of that noise clears in a single day. That is exactly why the window needs to run several weeks rather than a few of them: long enough for those separate, mismatched clocks to average out against each other instead of all landing in the same direction by coincidence. Once you have two or more weekly averages, compare them. If this week's average and last week's average sit within a pound of each other, and you logged food honestly across both, the average of what you ate over that stretch is close to your real maintenance calories for that period. Nothing else needs calculating.
Logging accuracy matters more here than it usually does
The method is only as good as the number on the eating side of it. Eyeballing portions introduces error in both directions, and that error does not have to be large to bend the result: consistently underestimating oil, dressing and cooking fat by even a modest amount across three weeks quietly inflates the maintenance number you end up calculating, because the intake side of the equation was never quite right to begin with. A kitchen scale for the foods that are easy to misjudge by sight, nut butter, oil, cheese, dense baked goods, does more for this specific measurement than it does for everyday tracking, precisely because you are about to treat the average as a fact rather than an estimate.
You do not need to weigh everything forever. You need the two to four weeks you are using for this measurement to be the most accurate logging you do all year, because a systematic bias in the input becomes a systematic bias in the answer, and it will not show up as noise you can average away.
What throws the measurement off mid-window
The method only holds if the window is otherwise ordinary. Two changes wreck it in opposite directions. Ramp up training hard during the measurement weeks, adding a session or pushing your sets meaningfully harder than usual, and you have raised what you burn mid-measurement; the flat weight at the end now reflects a different, higher number than the same flat weight would have shown at your old training load. Get sick, travel, or cut training back for the same stretch and the opposite happens.
Alcohol is its own case, worth naming on its own because it hides well in a log. A few drinks add real calories on the intake side, but they also tend to arrive on nights that disrupt sleep and shift what gets eaten the next day, which makes an evening like that a compound variable rather than a clean data point. It does not have to be excluded from the measurement window. It does have to be logged honestly, the same as everything else, or the average quietly stops meaning what you think it means.
Either way, the average you calculate from a window with one of these in it is real. It just describes that particular combination of eating and activity, not a number you can carry unchanged into a different week. This is one more reason it pays to check your training log rather than trust how a week felt. A sense that "training was about the same" is not reliable evidence that your sets and effort actually held steady, and a log showing a real jump or drop in volume during your measurement window is the flag that whatever number you calculate needs a caveat attached to it.
Reading your own numbers
Pull up the last month in your log and find two consecutive weeks where the weight averages sit within a pound of each other. Average what you logged eating across those same two weeks. That number, not a formula's guess, is close to your real maintenance calories for the training and activity level you were carrying during that stretch.
If no two-week stretch in your last month holds that flat, the method has not failed. It means your weight has genuinely been trending, up or down, and the more useful number right now is the direction of that trend rather than a single maintenance figure pulled from a period that was never actually stable.
The honest limit
A measured maintenance number describes the past. It is not a prediction of the future. It reflects the training you were doing, the sleep you were getting and the conditions you were living in during those specific weeks, and research on how the body adjusts its own energy use in response to sustained changes in eating or activity shows the relationship between intake and weight does not stay fixed once either one changes for long enough. Gain muscle, lose it, change how much you sleep, or spend a season somewhere colder, and the number you measured stops applying and needs remeasuring, the same way a stale formula-based guess would.
It also has nothing to say about body composition. Two people can hold the same weight on the same intake with very different amounts of muscle and fat underneath it, and this method cannot see that difference at all. It answers one question, cleanly, and is worth using precisely because it does not pretend to answer the others.
Adonae tracks your food, training, body and watch data together, so your logged calories and your weekly weight average sit in the same place instead of two apps you have to cross-reference by hand. Free to start, works with no account. Get it for Android, or use the web app.
