How many calories should I eat to lose weight?

This is the entire calculation. You can do it on paper in two minutes, and it is the same sequence Biteline runs at the end of onboarding.

Step 1 — Resting burn, with Mifflin-St Jeor

Basal metabolic rate is roughly what you would burn lying still all day. The equation takes weight in kilograms, height in centimetres and age in years:

• Men: (10 × kg) + (6.25 × cm) − (5 × age) + 5
• Women: (10 × kg) + (6.25 × cm) − (5 × age) − 161

The two versions differ only by a constant. Biteline offers a “prefer not to say” option in onboarding, and averages the two constants for it, because picking one would misstate the result for half of those users.

Step 2 — Multiply by how active you are

BMR is resting burn only. Everything else you do in a day — walking, standing, fidgeting, training — is folded in as a single multiplier, which turns BMR into total daily energy expenditure, or maintenance. This is the step that decides most of your answer, and it gets its own section below.

Step 3 — Subtract the deficit

A kilogram of body fat holds approximately 7,700 kcal. That convention is what converts a weekly goal into a daily number:

• 0.25 kg a week → 0.25 × 7,700 ÷ 7 = 275 kcal a day
• 0.5 kg a week → 0.5 × 7,700 ÷ 7 = 550 kcal a day
• 0.75 kg a week → 0.75 × 7,700 ÷ 7 = 825 kcal a day

Step 4 — Apply the floor

If the arithmetic lands below a defensible minimum, the minimum wins. Biteline never issues a target under 1,500 kcal for men or 1,200 kcal for women. If the equation produces less, you get the floor and you lose weight more slowly than you asked to. That is deliberate, and the third worked example below is exactly that case.

Three real profiles through the same four steps. Read the column that looks most like you, then redo it with your own numbers.

A — the ordinary case

(10 × 78) + (6.25 × 168) − (5 × 34) − 161 = 1,499 kcal at rest. Multiplied by 1.2 for a desk job, maintenance is about 1,799 kcal. Take off 550 and the target is 1,249 kcal a day.

Worth noticing: that is only 49 kcal above the floor. A woman of this size and activity level is close to the point where half a kilogram a week stops being a comfortable ask. Dropping to 0.35 kg a week gives her about 1,414 kcal instead — a target she is far more likely to still be eating in March.

B — the roomier case

(10 × 95) + (6.25 × 182) − (5 × 41) + 5 = 1,888 kcal at rest. He trains three to five days a week, so ×1.55 puts maintenance near 2,926 kcal, and the same 550 kcal deficit leaves 2,376 kcal a day. The same weekly pace, more than a thousand extra calories to work with — that gap is body size and activity, not willpower.

C — the case where the floor overrides you

(10 × 61) + (6.25 × 158) − (5 × 62) − 161 = 1,127 kcal at rest, ×1.2 for maintenance of about 1,352 kcal. She asks for 0.8 kg a week, the fastest pace the app offers, which is a deficit of 880 kcal — and 1,352 − 880 is 472. The app does not show that number. It shows 1,200.

Biteline offers a pace between 0.1 and 0.8 kg a week and defaults to 0.5. That default is not arbitrary, and neither is the ceiling.

The case for a moderate deficit is mostly practical. A 550 kcal gap is a smaller breakfast or a walk, and it survives a bad week; a 1,000 kcal gap is a project, and projects get abandoned. Weight loss is decided by the deficit you actually keep to over months, not the one you set in January. A slower plan that runs for twenty weeks beats a faster one that runs for three.

The case for a ceiling is different. Below a certain intake it becomes genuinely difficult to get enough protein and micronutrients out of normal food, and the arithmetic that produced the number stops being the only thing that matters. That is why the pace stops at 0.8 kg a week and why the calorie floors exist underneath it. We would rather cap the slider than hand out a deficit we could not defend.

Two things the 7,700 figure does not tell you

• It is a convention, not a constant. It approximates the energy in a kilogram of body fat, and what you lose is never purely fat. Treat it as the arithmetic everyone uses to turn a weekly goal into a daily one, not as a physical law about your body.
• The first week will not match it. Cutting calories usually cuts carbohydrate, glycogen carries water with it, and the scale drops further and faster than 550 kcal a day can account for. That is not fat, and it does not continue. Judge the plan from week three onwards.

Sex, age, height and weight are facts. The activity multiplier is a judgement you make about yourself, and it moves the answer more than anything else in the calculation.

Take example B above. At 1,888 kcal of resting burn, sedentary gives him a maintenance of 2,265 kcal and athlete gives him 3,586 — a difference of 1,321 kcal a day from one tap in a quiz. That single choice is worth nearly two and a half times his entire weight-loss deficit.

How to pick it honestly

• Rate your week, not your best week. The question is what a normal seven days looks like, including the ones where you cancelled the gym.
• When you are between two levels, take the lower one. An inflated multiplier produces a target that looks generous and does not move the scale, and you will spend a month blaming yourself rather than the input.
• Do not add your workouts on top. The multiplier already includes them. Logging an hour of exercise and then eating the calories back double-counts it — and the burn figure your watch reports for that hour is itself an estimate with a wide error bar.
• A physical job outranks a gym habit. Eight hours on your feet moves the multiplier more than four gym sessions do.

Mifflin-St Jeor is the standard estimator, and Biteline uses it rather than the older Harris-Benedict equation. Both are estimates of the same quantity. Neither measures your metabolism, and it is worth being precise about what they miss.

• It cannot see body composition. The equation takes total weight. Two people who weigh 80 kg at the same height and age get the same number, whether one of them is carrying considerably more muscle than the other. Muscle and fat do not burn energy at the same rate, so the estimate cannot be equally right for both.
• It cannot see the rest of your day. Everyday movement — pacing on calls, taking stairs, being restless — varies a great deal between people, and a five-step multiplier is a very coarse way to capture it.
• It is a population average applied to one person. Equations of this kind are fitted to groups. Any individual can sit above or below the line their inputs predict, and nothing in the formula reveals which one you are.
• It says nothing about how to spend the calories. The target is a total. How you divide it between protein, carbs and fat is a separate question, and a smaller one than most apps imply.

None of that makes the number useless. It makes it a starting hypothesis — good enough to begin with, and specific enough that the scale can tell you within a month how far off it was.

This is the part that decides whether any of the arithmetic above matters, and it is the reason to be suspicious of a target quoted to the calorie.

A 0.5 kg per week deficit is 550 kcal a day. A restaurant plate of pasta can be 550 kcal away from another restaurant plate of pasta. When we tested three AI vision models on thirty meal photographs, the highest and lowest estimate for the same plate differed by a median of 49 per cent — and for a 1,100 kcal dinner, 49 per cent is almost exactly one day’s deficit. Hand-estimated portions and database entries carry error of a similar order.

The practical conclusion is not that tracking is pointless. It is that precision is the wrong thing to chase, and consistency is the right one:

• Use the same method every day. A log that is consistently 15 per cent low is far more useful than one that is perfect on Monday and absent on Saturday, because a steady bias cancels out the moment you compare weeks against the scale.
• Log the meals you would rather not. The unlogged handful of crisps is the single largest source of error in most people’s diaries, and it only ever points one way.
• Spend your effort on portion size. It is the biggest lever on the number and the one you can judge yourself, standing in front of the plate. There are reliable ways to estimate a portion without a scale.
• Read the target as a band. If the app says 1,900 kcal, aim for it and treat anything between roughly 1,800 and 2,000 as having hit it. Chasing the last twenty calories is measuring noise.

The equation gives you a starting point. Your own weight trend gives you the real answer, and it takes about a month to read.

1. Weigh yourself the same way. Same time of day, ideally on waking, before eating. Daily if you can bear it, otherwise a fixed morning each week.
2. Ignore the first week. Water and glycogen dominate it, in both directions.
3. Compare week four against week two. Two averages, not two single readings — day-to-day weight moves by a kilogram or more on food, salt and sleep alone.
4. Convert the gap back into calories. Multiply the difference between the loss you got and the loss you planned by 7,700 and divide by 7. That is the daily correction.

An example. You aimed for 0.5 kg a week and averaged 0.25. The gap is 0.25 kg a week, which is 0.25 × 7,700 ÷ 7 ≈ 275 kcal a day. Either your maintenance is about 275 kcal lower than the equation estimated, or you are eating about 275 kcal a day more than you are logging. Both are common and the fix is the same: bring the target down by roughly that much, or add the movement to cover it, and give it another three weeks.

Do this twice and you have something the formula could never give you — a maintenance figure measured on your own body rather than predicted from four inputs. Your real maintenance is simply whatever intake holds your weight steady, and only the scale can tell you what that is.

If the weight has not moved at all after a month of consistent logging, resist the urge to cut deeper straight away. Check the multiplier first, and check what is missing from the diary. A deficit that exists only on paper is not a deficit.

Onboarding asks for the six things the calculation needs — sex, birth date, height, current weight, activity level and goal — plus a target weight and a pace. It then runs exactly the equation above: resting burn, activity multiplier, deficit, floor.

Three details worth knowing, because they are choices rather than arithmetic:

• The pace slider stops at 0.8 kg a week, and the calorie floors sit underneath it. You cannot talk the app into a target it should not give you.
• The target recalculates whenever an input changes — a new weight, a different activity level, a changed goal. A target set at 95 kg is the wrong target at 88 kg, and it should not be left standing.
• The macro split is anchored to your target weight rather than your current one, so the protein figure describes the body you are working towards instead of the one you are leaving.

After that it is a food diary: photograph the plate, correct anything the estimate got wrong, and watch the day fill against the ring. You can read the rest of how Biteline works on the home page.