You just finished a run that felt great. You glance down at your watch: heart rate looks normal, pace is solid, and then that calorie number pops up — and you feel instantly justified in eating that extra snack.

But what if I told you that number you're trusting could be off by as much as 25%?

And it's not just calories. What about heart rate? Sleep? Blood oxygen? How much of what's on your wrist every day is actually real?

The answer might surprise you: some of it is worth trusting, and some of it is little more than an educated guess.

Heart Rate: The Most Reliable Metric — With One Catch

Of everything a wearable measures, heart rate is widely considered the most accurate. One study of hospitalized patients found that the Fitbit Sense 2 measured heart rate within -0.3 bpm of a medical reference device, with a correlation of 0.998.

But "accurate" comes with an important caveat: it's reliable at rest, less so during exercise. A study comparing four popular watches found that resting heart rate errors were under 3 bpm, but during high-intensity exercise, errors ballooned to 8–16 bpm. Optical sensors work by shining light through your skin to detect blood flow changes — and when your wrist is moving and sweating, that signal gets muddy.

The verdict: For tracking resting heart rate trends — like noticing your resting HR slowly dropping as your fitness improves — it's more than good enough. But if you're doing precise heart-rate-zone training, a chest strap is still the gold standard.

Step Count: Fine for Daily Life, Less So for Running

Step counting is one of the oldest metrics around. While walking, accuracy is high — most modern devices stay within 5%. One review noted that step count is among the most frequently validated metrics.

But switch to running, and accuracy drops. The same research found that step count variability increases significantly during running. And devices don't always agree with each other — one study found Fitbit consistently overestimated steps.

The verdict: Use it to answer "did I move today?" Don't use it to precisely count your running strides.

Calorie Burn: The Most Overestimated Metric of All

This is where things get really shaky.

Study after study has reached the same conclusion: smartwatch calorie estimates typically have an error rate of 15%–25%, and in the worst cases, 30%–80%.

Even more important: the higher your body fat percentage, the less accurate the calorie data becomes. A PLOS One study found that error rates grew as participants' body fat increased.

Why? Because your watch isn't actually measuring how much energy you burned. It's estimating it based on heart rate, wrist motion, your height, and your weight. Every person's metabolism is different — two people with the same heart rate and movement can burn very different amounts of energy.

The verdict: If you're using calorie numbers to manage your weight, they're likely giving you false confidence. Researchers suggest treating the calorie number as an upper limit — your actual burn is probably lower.

Sleep: Total Duration Is Okay, Sleep Stages Are Not

Sleep tracking is one of the most popular wearable features — and one of the most questionable in terms of accuracy.

The good news: total sleep duration estimates are "moderately accurate" in healthy adults. One study found Fitbit's total sleep time was within -0.1 hours of a reference device — but the range of agreement was wide (±4 hours).

The bad news: sleep stages are largely unreliable. In the same study, Fitbit overestimated REM sleep by 29.4 minutes, with a correlation of just 0.41. Another evaluation of eight devices found that wrist-worn devices achieved sleep-stage accuracy of only 55%–72%, compared to 83%+ for chest-worn devices.

There's another issue worth noting: most wrist-worn devices systematically overestimate sleep time while underestimating night wakings. In other words, you might feel like you tossed and turned all night, while your watch reports a solid night's rest.

The verdict: Watch your sleep duration trends and consistency — those are useful. But don't take "deep sleep" and "REM" numbers too seriously. And don't stress because your watch says you didn't get enough deep sleep — it probably didn't measure it accurately anyway.

Blood Oxygen: Looks Advanced, But It's the Shakiest of All

Blood oxygen monitoring is a headline feature on many newer smartwatches — but it may be the least trustworthy metric of the bunch.

Medical pulse oximeters use a "transmission" method: light passes through your finger, emitted on one side and received on the other. A watch can only use "reflectance" — light goes in, bounces back through tissue. The wrist has deeper blood vessels and more complex tissue, which means far more signal interference.

The research is blunt: smartwatch blood oxygen measurements show large errors and significant variation between brands. When blood oxygen is genuinely low, watches tend to artificially read high. One clinical study went so far as to recommend that SpO2 data be treated with caution and not used for clinical decision-making.

The verdict: For healthy people at normal levels, a rough trend is about all you get. If you genuinely need to monitor blood oxygen — say, with a respiratory condition — a fingertip medical pulse oximeter is what you should trust.

So How Should You Actually Use This Data?

A practical approach: sort your metrics into three tiers by trustworthiness.

Tier 1: Use these to adjust your training

• Resting heart rate trend (a steady decline = improving fitness)

• Weekly training volume (time/distance)

• Consistency of sleep duration

Tier 2: Treat as hints, not truth

• Overnight HRV trends

• Composite scores like "recovery score" or "body battery"

• Stress index

Tier 3: Barely worth believing

• Precise calorie numbers

• Precise sleep stages

• Wrist-based blood oxygen as a health indicator

How to Choose the Right Device for You

If heart rate accuracy is your priority, Apple Watch has performed strongly in independent testing, with heart rate error rates under 1% (about 1.4 bpm). If you want long battery life and deep training analytics, Garmin is a common choice among runners and cyclists. If you prefer minimal, sleep-friendly wear, a ring-style device like the Oura Ring has a comfort advantage — though its sleep-stage accuracy is still only "moderate."

But no matter which you choose, remember one principle: you don't need every second of data to be accurate — you need the trend to move in the right direction. Resting heart rate slowly dropping, sleep duration slowly stabilizing, weekly training volume increasing sensibly — those trends matter far more than any single "precise" number.

What You Wear Determines Whether You'll Actually Keep It On

Here's something easy to overlook: the most accurate device in the world is useless if you don't want to wear it.

And how comfortable a watch or band feels depends a lot on what you're wearing. A tight long sleeve can pin the watch against your wrist and cause chafing on a run. A loose cuff lets the watch slide around — and when an optical sensor shifts, it leaks light and the data suffers.

For workouts, look for training tops with stretchy, non-binding cuffs, or long sleeves with thumbholes that keep the watch sitting securely in place. And if you wear a band on your ankle (some people do, to avoid wrist-motion interference), you'll want tights or shorts that aren't tight at the ankle.

The Bottom Line

Your smartwatch isn't lying to you — but it isn't a medical-grade instrument either. Heart rate trends, step count, and sleep duration — those you can trust. Calories, sleep stages, and blood oxygen — take them as rough references, not facts. And most importantly: let the data help you get better, not let the numbers run your life.