Every TDEE calculator on the internet — including this one — is an estimate, not a measurement. The honest accuracy ceiling is about ±10% of your true TDEE, which works out to roughly 200–300 kcal for a typical adult. That's not a flaw to hide; it's a structural feature of trying to predict a complex biological number from four inputs. This page lays out where the error comes from, what the studies actually say, and how to converge on your real TDEE in 4–6 weeks of consistent eating.
Where the ±10% comes from
The error isn't random — it's structural. Four specific sources, in roughly decreasing order of impact:
Activity multiplier guesswork. The activity tier you choose multiplies BMR by a coefficient between 1.2 and 1.9 — a 58% range. Real-world activity rarely aligns perfectly with the tier description. Most people overestimate, some underestimate, and almost nobody hits the multiplier dead-on.
NEAT variation (the hidden component). Non-exercise activity thermogenesis — fidgeting, walking, standing, posture changes — varies by hundreds of calories between people of identical demographics. Levine's NEAT studies documented an ~800 kcal/day spread between high-NEAT and low-NEAT individuals of the same size.
BMR formula error. Even the best formula (Mifflin-St Jeor) sits within ±10% of measured BMR for ~80% of healthy adults. The other 20% sit outside that range, mostly due to differences in lean mass that the formula can't see.
Self-reported inputs. People misjudge their own weight, height, and especially activity level. A study of self-reporters consistently shows over-estimation of training intensity and frequency.
What ±10% looks like in real numbers
Take a typical 30-year-old woman: 5'5" (165 cm), 145 lb (66 kg), lightly active. Her calculated TDEE is about 1,909 kcal. A ±10% confidence interval means her true TDEE is between 1,718 and 2,100 kcal. That's a 382 kcal/day window.
For weight loss, that's enormous. The difference between eating at 1,718 (her low estimate) and 2,100 (her high estimate) is roughly the difference between maintaining and losing 0.75 lb/week. If you treat the calculator output as an exact prescription, you'll get one of three outcomes: you'll lose faster than expected, you'll lose slower than expected, or you'll match. Roughly a third of people land in each bucket.
The solution isn't a more precise calculator. It's calibrating to your own body.
What the research actually shows
Frankenfield's 2005 comparison study found Mifflin-St Jeor predicted measured BMR within 10% in 82% of non-obese subjects and 70% of obese subjects. For TDEE — which includes the activity multiplier — accuracy drops further because that multiplier is a population-average shortcut, not an individual prediction.
Doubly-labeled water studies (the gold standard for measuring TDEE in free-living humans) consistently show that formula-derived TDEE estimates deviate by 10–20% from measured TDEE on an individual basis, with the larger errors concentrated in people who overestimate or underestimate their activity tier.
Field note from Sukie
The 14-day calibration protocol I use
Here's the exact process I've used on myself, my partner, and the two friends I've helped through fat-loss phases:
Day 0: Calculate TDEE from the formula. Pick your goal target (e.g., −20% for a moderate cut).
Days 1–14: Eat at your target every day. Be honest with tracking — weigh food on a kitchen scale, log everything including oils and condiments. Weigh yourself every morning at the same time, in the same clothes (preferably none), after the bathroom, before food or water.
Day 8: Average days 1–7. Don't react to the number yet.
Day 14: Average days 8–14. Compare to the day 1–7 average. The trend between those two averages is your real signal.
Day 15: Adjust. If you cut at a 20% deficit and lost about what you'd expect (0.5–1.0% body weight per week), you're calibrated — keep going. If you lost less, drop calories by 100 kcal and run another 2 weeks. If you lost much more and feel run-down, add 100 kcal back.
After three cycles (six weeks), your real TDEE is dialed in to within 50 kcal. That's better than any formula will ever give you.
— Sukie's actual protocol
Why daily weigh-ins (not weekly) are the right move
A single weigh-in can swing 2–3 lb in either direction from water, sodium, hormones, glycogen, or simply bowel contents. Weekly weigh-ins are too noisy — one bad day can derail your read of an entire week.
The fix is daily weigh-ins averaged over 7 days. Day-to-day noise washes out in the average. The trend across two consecutive 7-day averages is much more stable and much more informative than any single number. This is also the only weighing approach that doesn't make people miserable, because you stop treating a 1.5 lb morning bump as a moral judgment.
What a real calibration looks like in numbers
Hypothetical reader: 28F, 145 lb, 5'5". Calculator says TDEE = 1,909, target = 1,527 for a 20% cut.
Week
Avg daily weight
Δ from prev week
What it means
Week 1
145.4 lb
+0.4 lb
Mostly water/glycogen shift from new diet. Don't panic.
Week 2
144.1 lb
−1.3 lb
Real loss starting. On pace.
Week 3
143.5 lb
−0.6 lb
Slowing. Within expected range.
Week 4
143.2 lb
−0.3 lb
Stalled? Or normal? Compare 2-week averages.
W3+W4 avg vs W1+W2 avg
−1.3 lb total
−0.65 lb/wk
Real signal: cut is working at ~0.65 lb/wk. Maybe drop 100 kcal to hit 0.8 lb/wk if desired.
When the formula will be unusually wrong for you
Some people are outliers, and the formula will miss them by more than 10%. The patterns I've seen most often:
Muscular men (or women) with low body fat — Mifflin underestimates. Switch to Katch-McArdle if you have a reliable body fat number.
Formerly obese individuals who've dieted down significantly — long-term adaptive thermogenesis suppresses TDEE 5–15% below where the formula predicts. The Biggest Loser follow-up study famously documented this.
People with thyroid disorders — hypothyroidism can suppress BMR 5–10%; treated hyperthyroidism can be slightly elevated.
Very young or very old adults — Mifflin-St Jeor is validated for ages 18–78. Outside that range, accuracy decreases.
For these populations, the 2-week calibration becomes even more important. The formula gets you in the ballpark; your own data finishes the job.
What the validation research actually says
When people argue about whether TDEE calculators 'work,' they're usually arguing past each other because they're using different definitions of accurate. The published validation literature is pretty consistent if you read it carefully.
For the BMR portion: Frankenfield's 2005 meta-comparison (Journal of the American Dietetic Association) tested four predictive equations against indirect calorimetry across non-obese adults. Mifflin-St Jeor predicted within 10% of measured BMR for roughly 82% of subjects. That's the often-cited '±10%' figure. The remaining 18% were off by more than 10% — almost always in the direction of formula overestimation, mostly because the subject was at the low end of typical lean mass for their body size.
For the TDEE portion (the bigger and harder problem), the activity multipliers are essentially population averages and carry their own error. The doubly-labeled water (DLW) literature, which measures real-world TDEE in free-living adults, shows that activity factor groupings (sedentary/light/moderate/active/very active) explain about 60–70% of inter-individual variance in TDEE at a given body size and age. The other 30–40% is NEAT variance plus exercise intensity variance plus measurement noise. A formula cannot recover that signal — only your own body weighed over time can.
A practical reading of the literature: the calculator's BMR estimate is within ±10% for most adults. The TDEE estimate compounds an additional ±5–8% from activity multiplier error. Combined error: typically ±10%, occasionally ±15%, rarely ±20% or worse. That's accurate enough to set a starting target, not accurate enough to skip the 2-week verification.
Why the same person gets a different number on every calculator
Plug your stats into five online TDEE calculators and you'll often see a 300–500 kcal spread across the results. That spread isn't because four of them are wrong and one is right. It reflects three legitimate choices each calculator made independently.
First, formula choice. Some calculators default to Mifflin-St Jeor, some still use Harris-Benedict (which runs 80–150 kcal higher), some use Katch-McArdle when body fat is provided. None of these is 'wrong' for the average user — they're different baselines with known accuracy tradeoffs.
Second, activity multiplier definitions. The labels 'lightly active' or 'moderately active' aren't standardized. Site A's 'moderately active' is 1.55; site B's is 1.5; site C's is 1.6. Same description, different math. This single inconsistency drives most of the spread between calculator outputs for the same user.
Third, rounding and intermediate calculation order. Some calculators round BMR before multiplying by the activity factor; some carry the unrounded BMR through the calculation. The two approaches can differ by 10–30 kcal on the final number.
The takeaway: don't treat any single calculator's output as 'the right number.' Treat it as a starting estimate with an inherent ±10% uncertainty band, and verify against your real weight trend over 2 weeks. Our calculator on this site uses Mifflin-St Jeor as the default (the most evidence-supported), shows you the math in the 'Show the math' panel, and gives you a calibration protocol — but even our number is an estimate until you've tested it against the scale.
What's the most accurate way to measure TDEE without a formula?+
Doubly-labeled water (DLW) is the gold standard but is expensive (~$500+) and only available through research labs. The practical alternative is the eat-and-weigh calibration: eat at a fixed target for 2–4 weeks while weighing daily, then back-calculate from your weight trend.
Are wearable trackers more accurate than the formula?+
Generally no — wrist-worn trackers tend to overestimate active calorie burn by 15–40%. Chest-strap heart rate monitors do better but still struggle with strength training. For TDEE, use the formula plus your real-world weight trend.
Does the calculator account for muscle mass?+
Mifflin-St Jeor doesn't directly. If you have a reliable body fat percentage and enter it, our calculator switches to Katch-McArdle, which calculates from lean body mass.
Why does my TDEE estimate differ between websites?+
Different formulas, different activity multiplier definitions, or different defaults. Most of the differences are within ±5%. Ours uses Mifflin-St Jeor with widely-used Harris-Benedict-style activity factors (1.2 to 1.9).
Can I trust the calculator if I'm new to fitness?+
Yes — as a starting point. Calculate it, eat at the target, weigh in daily, and adjust after 2 weeks. The formula gets you within striking distance; reality does the fine-tuning.
Should I retest my TDEE periodically?+
Yes. Recalculate every 10 lb of weight change or every 3–4 months of consistent eating, whichever comes first. Your TDEE drifts as you change size and as your habits change.