Run your stats through two macro calculators and you will often get two different calorie targets for the exact same body. A 27-year-old woman at 165 cm and 68 kg might read 1,415 calories on one site and close to 1,500 on another before she has touched a single other input. Neither site is broken. They are using different equations to estimate the same thing, and the gap between those equations has a documented history that goes back more than a hundred years.
That gap trips people up regularly. You pick a calculator, you trust the number, and then a friend or a coach runs a different one and gets a figure that is off by a couple hundred calories. It feels like one of them has to be wrong. Usually neither is. They just made different choices at the very first step, before macros ever enter the picture.
Where your calorie target actually comes from
Every macro plan starts with a single estimate: how much energy your body burns. That number gets divided into protein, carbohydrate, and fat later, but the division is downstream. If the calorie figure at the top is off, every gram target under it shifts with it.
The figure at the top is your basal metabolic rate, usually written as BMR and sometimes reported as resting metabolic rate. It is the energy you spend keeping the lights on at rest: breathing, circulation, cell repair, holding your body temperature steady. For most people that baseline accounts for somewhere around 60 to 70 percent of everything they burn in a day. You can't pin it down exactly without a lab, a ventilated hood, and a stretch of lying perfectly still, so calculators estimate it from the things you can measure at home: sex, age, height, and weight.
That estimation step is where calculators quietly part ways. There is no single agreed equation. The ones in common use were built from different groups of people, in different decades, for different reasons, and they don't return the same answer.
Three equations, three answers
The oldest one still in circulation is Harris-Benedict, published in 1919 out of the Carnegie Institution. It was the standard for decades and it still turns up in fitness tools today. In 1984, Roza and Shizgal reworked the coefficients against a larger sample and produced a revised version that runs a little lower than the original. Both take the same four inputs.
Then in 1990 came Mifflin-St Jeor, published in the American Journal of Clinical Nutrition. It was fit to a more modern population and was designed specifically to predict resting energy in healthy adults more reliably than the older equations. It uses the same sex, age, height, and weight inputs, which makes it a clean drop-in replacement, and it tends to land a bit below Harris-Benedict.
Here is what those three look like for the same person, a 27-year-old woman at 165 cm and 68 kg:
| Equation | First published | BMR estimate |
|---|---|---|
| Original Harris-Benedict | 1919 | ~1,484 kcal |
| Revised Harris-Benedict (Roza–Shizgal) | 1984 | ~1,471 kcal |
| Mifflin-St Jeor | 1990 | ~1,415 kcal |
That is a spread of about 69 calories at rest, from identical inputs, purely because the equations were fit to different data. It isn't a rounding artifact. It's three honest attempts to model the same body that happen to disagree.
There is a fourth option you will see on the more detailed calculators: Katch-McArdle. It ignores the standard inputs and works from lean body mass instead, which means you have to know your body-fat percentage to use it. The idea is sound, since muscle is more metabolically active than fat, so two people at the same weight but different builds should not get the same BMR. The catch is that most people are guessing at their body-fat number, and a guess there feeds straight into the result. For anyone who does not have a reliable body-composition reading, Katch-McArdle trades one estimate for another rather than removing it.
Why Mifflin-St Jeor became the default
If you look at what modern calculators reach for first, it is almost always Mifflin-St Jeor. That is not fashion. In 2005 the Academy of Nutrition and Dietetics, then the American Dietetic Association, published a systematic review by Frankenfield, Roth-Yousey, and Compher in its journal. The authors compared the common predictive equations against resting metabolic rate that had actually been measured in a lab, across both leaner and heavier adults.
Mifflin-St Jeor came out on top. It predicted measured resting energy within 10 percent for more people than Harris-Benedict or the other equations in the comparison. That is the finding that pushed dietitians and, eventually, the calculators they influenced toward it as the general-population default. It also lines up with the table above: Harris-Benedict tends to run high, and running high means handing people a maintenance number that is a little too generous for the goal they came in with.
None of this makes Harris-Benedict useless. It made sense with the data available in 1919, and the revised version narrowed the gap. But when a calculator has to pick one equation to show by default, the evidence points at Mifflin-St Jeor, and that is the honest reason to prefer it over picking one because it returns a friendlier number.
The multiplier does more damage than the formula
Here is the part that surprises people who obsess over which BMR equation to trust. BMR is only the resting base. To get the number you actually eat against, the calculator multiplies it by an activity factor, and those factors are far cruder than the equations they scale.
The standard activity multipliers run from 1.2 for sedentary up to about 1.9 for extra active, in a handful of wide buckets. Take our example woman at her Mifflin-St Jeor BMR of 1,415. Call her lightly active and the multiplier of 1.375 puts her maintenance near 1,946 calories. Call her moderately active instead and the 1.55 multiplier puts it near 2,193. That single dropdown choice moved her target by roughly 247 calories, which is more than three times the 69-calorie spread across all three BMR formulas combined.
So the input people agonize over, which equation to use, matters less than the input they pick casually off a dropdown. And activity level is self-reported, fuzzy, and easy to inflate. Most people rate themselves a tier higher than their week actually supports, especially if they count a few gym sessions but sit for the other fifteen hours a day. If you want to worry about accuracy, the activity bucket deserves more of that worry than the formula does.
What the number can and can't tell you
Even the best-validated equation is a population average, and averages hide individuals. The same Frankenfield review that crowned Mifflin-St Jeor also quietly set its ceiling: within 10 percent for the majority, which means for roughly one person in five the estimate is off by more than 10 percent in one direction or the other. On a 1,415-calorie base, 10 percent is more than 140 calories. Your real metabolism could sit meaningfully above or below the figure any calculator hands you, and no set of four inputs will tell you which.
That isn't a reason to distrust the math. It's a reason to treat the output as a starting line rather than a verdict. The workable approach is to take the estimate, eat at it for two to three weeks, watch the trend on the scale rather than any single day, and adjust from what your own body actually does. The equation gives you a defensible place to begin; your results calibrate it from there. Anyone who tells you a formula can hand you your exact calorie need to the digit is selling more certainty than the science supports.
Once you know that, the useful thing a calculator can do is show its work: which equation it used, which multiplier you picked, and how the number moves when you change your mind about either. Macra runs the Mifflin-St Jeor formula in your browser and shows it alongside your result, so you can see exactly how your starting number was built before you begin adjusting away from it.