And what to do about it.
Every calorie tracking journey starts the same way: you Google "TDEE calculator," plug in your height, weight, age, and activity level, and get a number. Maybe 2,200 kcal/day. You eat that for two weeks, step on the scale, and nothing has changed. Or worse, you've gained.
The calculator wasn't necessarily broken. But it was doing something most people don't realise: guessing.
The most widely used formula in calorie calculators is Mifflin-St Jeor, published in 1990. It estimates your Basal Metabolic Rate (BMR) — the calories your body burns at complete rest — from four variables: weight, height, age, and sex.
To get your Total Daily Energy Expenditure (TDEE), calculators multiply this BMR by an "activity factor" — a number between 1.2 (sedentary) and 1.9 (very active) that's supposed to capture everything else you do in a day.
In a 2005 systematic review, Frankenfield, Roth-Yousey, and Compher compared the major prediction equations and concluded that Mifflin-St Jeor was the most reliable, predicting within 10% of measured resting metabolic rate for the largest proportion of subjects. The Academy of Nutrition and Dietetics subsequently adopted it as their recommended equation.
So it's the best we have. But "best among formulas" and "accurate for you" are very different claims.
Even in controlled research settings, Mifflin-St Jeor has a standard error of roughly 150–200 kcal/day at the group level. For individuals, the error can exceed 300 kcal/day. That same Frankenfield review found that while the equation predicted within 10% for most people, a meaningful minority fell outside that band.
To put 300 kcal in perspective: that's the difference between losing half a kilogram per week and maintaining your weight. It's a banana and a tablespoon of peanut butter. It's enough to completely stall your progress or make you think your diet is broken when it's actually the number that's wrong.
The formula uses four inputs. Your body has hundreds of variables that affect energy expenditure. Here are the ones that matter most:
NEAT is the energy you burn through fidgeting, posture, walking to the kitchen, tapping your foot — everything that isn't deliberate exercise or basic organ function. In 1999, Levine, Eberhardt, and Jensen published a study in Science that overfed subjects by 1,000 kcal/day for 8 weeks. Fat gain varied 10-fold between participants, and the primary explanatory variable was NEAT. Some people unconsciously ramped up their movement by nearly 700 kcal/day. Others barely changed.
No formula captures this. And NEAT doesn't just vary between people — it varies within the same person depending on sleep, stress, and how much they ate the day before.
The discrete categories — sedentary, lightly active, moderately active, very active — trace back to a 1985 FAO/WHO/UNU report based on limited population data. They impose rigid boundaries on what is a continuous, highly individual spectrum.
Are you "lightly active" or "moderately active"? The difference between those two multipliers (1.375 vs 1.55) applied to a BMR of 1,600 kcal means a TDEE gap of 280 kcal/day. Pick the wrong category and your target is off by more than a meal's worth of food.
Doubly labelled water studies — the gold standard for measuring free-living energy expenditure — consistently show that people overestimate exercise and underestimate sedentary time. The category you think describes you probably doesn't.
In 1986, Ravussin and colleagues used whole-room calorimetry on 177 subjects and found that fat-free mass, fat mass, age, and sex together explained about 80% of the variance in 24-hour energy expenditure. That leaves roughly 20% driven by genetics, hormones, gut microbiome composition, sympathetic nervous system activity, and other factors that no online calculator asks about.
Even if a formula nails your TDEE today, it will be wrong in six weeks if you're in a calorie deficit. The reason is metabolic adaptation — your body's response to sustained energy restriction.
In 1995, Leibel, Rosenbaum, and Hirsch published a landmark paper in the New England Journal of Medicine showing that a 10% reduction in body weight decreased TDEE by more than the loss of tissue alone could explain. The body actively downregulates energy expenditure — lowering BMR, reducing the thermic effect of food, and making physical activity more metabolically efficient.
A 2010 review by Rosenbaum and Leibel in the International Journal of Obesity estimated that weight-reduced individuals burn roughly 300–400 kcal/day less than their new body composition would predict. This adaptation involves the thyroid axis, leptin signalling, and sympathetic nervous system changes.
The most striking evidence came from Fothergill and colleagues (2016), who followed 14 contestants from "The Biggest Loser" for six years. Despite most having regained significant weight, their resting metabolic rate was still about 500 kcal/day below expected. The adaptation persisted years after the diet ended.
A static calculator recalculated at your new weight would still be wrong, because it assumes normal metabolic rate for someone of that size. Metabolic adaptation means you're not that person anymore — at least not metabolically.
If formulas are unreliable, what's the alternative? Measurement.
The concept is simple: instead of predicting your TDEE from a formula, calculate it from what actually happened. If you know approximately how much you ate over a week and how your weight changed, basic physics gives you the answer:
If you averaged 2,100 kcal/day and lost 0.3 kg over 7 days, your implied TDEE is about 2,430 kcal/day. No formula needed. No activity multiplier. No guessing about your fidgeting habits.
This is what Clawrie does. Instead of calculating your TDEE once and hoping it's right, it continuously recalculates from your logged intake and real weight data using a sliding 7-day window with exponential smoothing. The estimate incorporates metabolic adaptation automatically, because it's measuring what your body is actually doing — not what an equation thinks it should be doing.
In the first week, Clawrie uses Mifflin-St Jeor as a starting point. But as data accumulates, the adaptive estimate takes over and the formula fades into irrelevance. Within 2–3 weeks of consistent tracking, the algorithm typically converges to within a useful range of your true TDEE — including any metabolic adaptation effects.
TDEE calculators are useful as a rough starting point. Mifflin-St Jeor is the best of the available formulas, and for many people it gets reasonably close. But "reasonably close" still means an error margin that can make or break your results.
The real issue isn't the formula — it's the assumption that your metabolism is a static number that can be computed once. It's not. It changes with your diet, your activity, your sleep, your stress, and your hormones. The only way to track a moving target is to keep measuring it.
References: Mifflin et al. (1990), Am J Clin Nutr. Frankenfield, Roth-Yousey & Compher (2005), J Am Diet Assoc. Ravussin et al. (1986), J Clin Invest. Levine, Eberhardt & Jensen (1999), Science. Leibel, Rosenbaum & Hirsch (1995), NEJM. Rosenbaum & Leibel (2010), Int J Obes. Fothergill et al. (2016), Obesity.