Your calorie calculator gave you a number. Now what?

The number a calorie calculator gives you is a starting estimate, not a measurement. Even the best-tested equation for resting energy lands within 10% of the measured value for only about four people in five³, and the activity level you pick moves it further: in the example below, each step is 315 kcal. So use it like this: eat to it for 14 days, log your food and your morning weight, compare the two weekly averages, and only then adjust by 100 to 200 kcal. Not sooner, and not by more.

Where the number comes from

Every calculator does two things. First it estimates your basal metabolic rate (BMR), what your body burns at complete rest, from sex, age, height and weight. Then it scales that up for your activity to get your total daily energy expenditure, your maintenance calories. Your goal (lose, maintain, gain) adds or takes calories from there.

Here's an example person. He's made up, not a real case: a man, 35, 178 cm (5 ft 10 in), 82 kg (181 lb), desk job, trains 2 to 3 times a week.

Two calculators, two answers

Calculators don't all do the same sum. They use different BMR equations and different activity scales. Here is the same man run through two well-known equations:

Which is closer? A systematic review of the four most used equations found that the newer one came within 10% of measured resting energy in more people than the others, the older one included¹. Its authors measured 498 people and reported that the original 1919 version of the older equation overestimated by 5% on average². In a later study of 337 adults, the newer equation was within 10% for 82% of them: 87% of people without obesity and 75% of people with obesity³. Roughly one estimate in five misses by more than 10%. The review says it plainly: for an individual, the error can be large¹.

For the example we round to 1,800 kcal, between the two.

10% of 1,800 is 180 kcal. Scaled up for his activity (light, see below), that's about 248 kcal a day either way. So from the equation alone, our man's real maintenance could sit anywhere from 2,228 to 2,723 kcal. And he hasn't picked an activity level yet.

The activity level moves the number more than the equation

Many calculators offer five levels. Here's the BMR of 1,800 kcal at each:

Each step adds 315 kcal. The lowest and highest levels are 1,260 kcal apart. One click in the dropdown moves your result further than the equation's own error, which for this man is about 248 kcal.

These levels are a convention that calculators share. We couldn't trace where they first came from. Measured data don't line up neatly with them. The lowest daily burn ever measured comes from people who cannot walk⁴, and the FAO scale puts a typical desk-bound adult well above a calculator's lowest level⁵. So a calculator's "sedentary" sits low, and its "moderate" isn't a measured figure.

It's easy to pick a level above the one you live at. In one study, 10 people with obesity who believed they couldn't lose weight on a diet overstated their activity by 51% on average (with a very wide spread) and understated their food by 47%⁶. That's a small, specific group and it doesn't speak for everyone, but it shows which way the mistake can go. Our rule is simple: if you're torn between two levels, take the lower one. Our man goes in as light, 2,475 kcal, not moderate.

You'll find advice online to eat at your BMR if you don't know your activity. We wouldn't give it. BMR is what a body burns lying still all day. If you really don't know, take the lowest level, and know that it's a cautious start: most adults who walk around burn more⁵. If your target then comes out very low, see the section on when to see a doctor before you start.

From maintenance to a target

Losing. Take off 500 kcal: 2,475 − 500 = 1,975 kcal a day. That's a 20.2% deficit. Deficits of 500 or 750 kcal a day are what the American guideline for treating overweight and obesity uses⁷. It was written for people in treatment under supervision, so 500 here is an example, not a prescription.

By the old rule of 3,500 kcal per pound (about 7,700 kcal per kg): 500 × 7 / 7,700 = 0.45 kg a week, about 1 lb, or 0.55% of 82 kg. The CDC calls 1 to 2 lb (0.45 to 0.9 kg) a week a gradual pace⁸. The rule is crude, though. Checked against real experiments, it overestimates weight loss⁹, and the body adapts, so weight drifts towards a new balance slowly, with a half-time of about a year in Hall's model¹⁰. Read 0.45 kg as a direction for the first weeks, not a promise for month six.

Maintaining. About 2,475 kcal.

Gaining. With a 10% surplus: 2,475 + 248 = 2,723 kcal. 10% is a convenient starting point, not a number from a study. A target of 0.25 to 0.5% of body weight a week works out at 0.2 to 0.4 kg for him, but that figure comes from a review written for bodybuilders¹⁶. We found no guideline on gain rate for the general population.

The first 14 days: reading the result

For two weeks you eat to the target, log everything you eat, and weigh yourself each morning. We cover how to read the scale without reacting to a single number in Your weight changes every day. Compare the average of the first seven mornings with the average of the next seven.

Say that after 14 days our man sees this. His logged intake averaged 1,950 kcal a day (a little under target, as happens in real life), and his weekly average weight fell by 0.15 kg a week, not 0.45.

What is he really burning? Work it backwards:

• At 7,700 kcal per kg: 0.15 × 7,700 / 7 = 165 kcal a day of deficit, so 1,950 + 165 = 2,115 kcal.
• At 2,380 kcal per kg, the average one study found for short-term ups and downs of 1 to 3 kg in people who weren't dieting¹¹: 0.15 × 2,380 / 7 = 51 kcal, so 2,001 kcal.

On paper his maintenance comes out between 2,001 and 2,115 kcal, which is 360 to 474 kcal below the 2,475 estimate. Two weeks of logs and weights can't pin it closer than that, and maybe not even that close. Both figures sit below the sedentary level (2,160 kcal).

We check the causes in this order:

1. The log. Logged intake usually comes in under real intake, and the gap grows the more someone eats¹². No logging method is more accurate than the others¹³. Researchers who defend food records still warn against treating logged calories as true intake¹⁴. So the log is a check, not a measurement, and it has to be complete. Any days missing? Cooking oil, drinks, the bites between meals? Food weighed, or guessed?
2. The activity level. It may be set higher than his real life⁶.
3. The equation. It may simply run high for him³.
4. Noise. In 46 adults eating as usual, the average change over two weeks was 0.26 kg, with a standard deviation of 1.2 kg¹¹. Two weeks is not much data.

Here's what we would do. If the log is complete and the food is weighed, drop the target 150 kcal below what he actually ate, to about 1,800 kcal, and wait another two weeks. A step of 100 to 200 kcal is coaching and app practice (one guide suggests about 100¹⁵), not a finding from a trial. We wouldn't cut the whole 360 to 474 kcal gap at once: two weeks of data are noisy, and a smaller step is easier to check.

There's another honest option. 0.15 kg a week is probably still weight loss. If another two weeks confirm it and the pace suits you, change nothing.

It works the other way too. If your weight drops faster than planned, you burn more than the estimate said. Same arithmetic, opposite sign. But don't judge by the first week: short-term changes on the scale are mostly not fat¹¹.

What you see and what to do

When this does not apply

• If you're under 18, pregnant or breastfeeding. These sums are for adults; the NIDDK planner, for one, says outright that it isn't for these groups¹⁷.
• If your metabolism is unusually low or you're very sedentary. NIDDK says its own planner then comes out too high¹⁷, and the same goes for any calculator.
• If you're older or belong to a group that studies have under-sampled. The review advises extra caution with the number¹.
• In the first week or two of a new diet. Short-term changes on the scale are mostly not fat¹¹, so those days don't give you a rate.
• If you don't log food. Without intake there's no back-calculation, only the weight trend.

When to see a doctor

On your own, don't go below 1,200 kcal a day if you're a woman or 1,500 if you're a man, whether with your first target or a later adjustment. These are the lowest intakes the obesity-treatment guideline prescribes under supervision⁷. They aren't a physiological minimum but the line below which you need a specialist. Below 1,000 kcal the NIDDK planner warns outright¹⁷.

If your BMI is under 18.5 and you want to lose weight, don't start a deficit; talk to a doctor.

The same goes if a doctor manages your eating or weight for a health condition, if you take medicines that affect appetite or weight, or diabetes medicines (insulin, sulfonylureas), or if you've had an eating disorder. Then count calories with a specialist, not alone.