Can you trust the calories-burned number on your watch?

A single glance at a watch face can hide three different numbers. Resting calories are what your body would spend doing nothing at all, just keeping your heart beating and your organs running. Active calories are the energy a device attributes to movement above that baseline: workouts, walking, fidgeting, anything that raises your effort above rest. Total calories are the two added together, the full day’s estimated burn.

The catch is that platforms do not all define these terms the same way. One app’s active calories might already include a small resting component, another might report only a raw total with no breakdown at all. The same word can mean a slightly different calculation depending on which app or device you are reading it from, which is worth knowing before you compare numbers across two sources.

No consumer device measures calorie burn directly. What it does is combine a handful of signals into a model and output an estimate. In plain terms, the inputs typically include your heart rate, your movement and step count, the workout type you selected or that the device detected, your pace or speed, and profile details like age, sex, height and weight. Those go into a device-specific formula that the manufacturer builds and refines on top.

Exactly how each platform weighs those inputs is proprietary, and we are not going to describe any specific vendor’s algorithm here unless it is something that vendor has documented themselves. What matters for you as a reader is simpler: more and better sensor data generally produces a steadier estimate, and an out-of-date profile or a poor sensor read degrades it no matter how good the underlying formula is.

A few sources of error show up across nearly every wearable, no matter the brand. Sensor quality and wrist fit matter: a loose band or unusual wrist anatomy can throw off a heart-rate-based estimate. Activity modality matters too. Strength training and other load-bearing work are notoriously harder to model than steady-state cardio, because the wrist barely moves while the body works hard, so a device that estimates well on a run can miss badly during a lifting session.

Individual physiology plays a role as well: two people of the same age, sex and weight can have real differences in resting metabolism and heart-rate response that a population-level formula cannot see. A stale profile, out-of-date weight or an incorrect setting quietly skews every estimate built on top of it. And because the models themselves are proprietary and undisclosed, there is no way for an outside party to audit exactly where a given number came from. On top of all that, some days simply have missing data: a skipped charge, a dropped sensor reading, or a workout logged without a heart-rate strap.

No. A device does not need to be a metabolic cart to be useful. Even an estimate with real error still carries a consistent signal: it can tell you which days were more active than others, how a training block changes your output over a few weeks, and whether the burn estimate broadly agrees with what your weight actually did over that same stretch. Error that is roughly consistent day to day is far more useful than error that is random, because a consistent bias still preserves the trend even when the absolute number is off.

The practical takeaway is to stop asking whether the exact number is right and start asking whether it moves the way your body actually moved. That second question is answerable from your own data, and it is the one worth paying attention to.

Here is how the three terms relate, with the caveat that exact platform definitions vary slightly from one source to another.

Some sources report all three. Others report only active calories and leave resting and total for you, or another app, to work out.

Because no device measures burn exactly, we do not present your wearable’s number as fact. The authenticated Calories Burned view instead places device burn, your logged intake and your maintenance estimate on one 90-day timeline, so you can see how the three relate rather than staring at any single figure in isolation.

Where a source reports a measured total, that measured value always wins and nothing is substituted for it. But some sources send active energy with no total and no resting record at all, depending on which permissions the phone granted and what that platform chooses to expose. In that specific gap, and only that gap, we rebuild a total by adding a profile-derived resting estimate to the device’s active figure. That reconstructed value is clearly labeled as estimated rather than passed off as something the device measured, and a day that already carries a measured resting reading is left alone rather than overwritten.

It is also worth keeping three related numbers distinct in your head, because they answer different questions: your device’s reported burn is what the wearable’s model estimated for that day; a formula-estimated resting burn is a profile-derived baseline built from the Mifflin-St Jeor equation; and the adaptive maintenance estimate is a separate figure derived from your logged intake and your weight trend over time. They inform each other, but they are not the same number, and none of them substitutes for the other two.

Rather than trusting or dismissing any single reading, triangulate. Compare your device’s estimate against your logged intake and a smoothed weight trend over multiple weeks, not one day. A single odd reading, a missed sensor read or an unusually intense session is normal and not worth chasing. What is worth paying attention to is a persistent mismatch that holds up over two or three weeks, since that is a sign your maintenance estimate or your logging needs a look rather than your device being simply wrong.

Read your maintenance calories against your intake and weight trend, and see how the full picture fits together on the calories in, calories out page. Treat the device number as one signal among several, not the final word.

Wellness Project reads calorie data directly from Fitbit, Apple Health, Health Connect, Oura and Withings. Those are the only sources we read calorie data from directly today.

Devices like Garmin, Samsung, Whoop, Zepp and Amazfit are not directly integrated for calorie data. Most of them commonly route through Apple Health or Health Connect on the phone side, so if yours does, its calorie data can still reach Wellness Project that way. We speak about that path generically rather than claiming a direct integration we do not have.

For how a single day’s total burn relates to your broader energy budget, see how TDEE is calculated and how it differs from a daily burn reading.

For readers who want to go deeper than a marketing page should, a couple of starting points:

• Shcherbina A, et al. Accuracy in Wrist-Worn, Sensor-Based Measurements of Heart Rate and Energy Expenditure in a Diverse Cohort. Journal of Personalized Medicine, 2017. A peer-reviewed validation study comparing several consumer wrist devices against a metabolic cart.
• Apple. Track daily activity with Apple Watch. Apple’s own published explanation of the move, exercise and stand figures behind its energy numbers.
• Fitbit Help. How does my Fitbit device calculate my daily activity?. Fitbit’s own published definition of how it estimates calories burned.