Fitness trackers use a combination of motion sensors, heart rate monitors, and increasingly sophisticated algorithms to estimate when you fall asleep, how long you stay asleep, and what sleep stages you cycle through during the night. No single sensor does the job alone. Your tracker is constantly collecting streams of data from your body and feeding them into software that pieces together a picture of your night.
Motion: The Foundation of Sleep Tracking
The most basic layer of sleep tracking relies on a tiny accelerometer inside your device, the same chip that counts your steps during the day. Most modern trackers use a tri-axial accelerometer, meaning it measures movement along three planes simultaneously. When your body is still for an extended period, the algorithm begins to suspect you’re asleep.
This approach, called actigraphy, has been used in clinical sleep research for decades. The tracker’s software analyzes multiple features of the accelerometer data, capturing signal variability and running it through a multi-step process that filters out noise. It’s not simply looking for “zero movement.” The algorithm distinguishes between the kind of stillness you’d have while reading in bed and the deeper, more sustained stillness of actual sleep, though this distinction is far from perfect. Quiet wakefulness is one of the hardest states for any motion-based system to correctly identify.
Heart Rate and Sleep Stages
To go beyond just detecting sleep versus wake, trackers use an optical heart rate sensor on the underside of the device. This sensor, called a photoplethysmography (PPG) sensor, shines green LED light into your skin and measures how much light bounces back. Because blood absorbs light differently depending on how fast it’s flowing, the sensor can read your pulse beat by beat.
Your heart behaves differently in each sleep stage, and this is what gives trackers their best window into what’s happening in your brain. During deep sleep (the restorative, non-REM stages), your heart rate drops and becomes very steady. During REM sleep, when most vivid dreaming happens, your heart rate rises again and becomes more variable, closer to what it looks like when you’re awake. The ratio between low-frequency and high-frequency components of your heart rhythm shifts predictably between these stages, giving the algorithm another signal to work with.
By combining the pattern of your heart rate variability with your movement data, the tracker estimates how much time you spent in light sleep, deep sleep, and REM sleep throughout the night. This is also how most devices detect brief awakenings you might not even remember.
Temperature and Your Internal Clock
Some newer trackers include a skin temperature sensor, and the data it collects is more meaningful than you might expect. Your body follows a predictable temperature cycle tied to your circadian rhythm. As you approach sleep, blood vessels in your hands and feet dilate, releasing heat from your core through your skin. This rise in skin temperature at your extremities actually helps trigger sleep onset.
Skin temperature at your wrist peaks during the night and drops in the morning as you wake. Trackers that measure this can use it as an additional signal for identifying when you fell asleep and how stable your sleep was. Temperature data also helps flag disruptions to your circadian rhythm. Interestingly, this pattern shifts with age: older adults show a weaker temperature cycle with the peak arriving over an hour earlier, which partly explains why sleep timing tends to shift earlier as people get older.
Blood Oxygen and Breathing Patterns
Many mid-range and premium trackers now include a pulse oximeter, a sensor that estimates blood oxygen saturation by shining red and infrared light through your skin. During normal sleep, your oxygen levels stay relatively stable. When breathing is disrupted, as happens with sleep apnea, oxygen levels dip repeatedly throughout the night.
Trackers flag these dips using something called an oxygen desaturation index, essentially counting how many times your blood oxygen drops by 3% or more per hour. Wearable ring devices like the Circul can capture beat-by-beat oxygen data overnight, while wrist-based trackers sample less frequently. This feature is not a medical diagnosis, but consistent patterns of oxygen dips can be an early signal worth discussing with a doctor.
Skin Conductance: A Newer Signal
A handful of devices, notably some Fitbit models, have added electrodermal activity (EDA) sensors that measure tiny changes in skin conductance caused by sweat gland activity. During sleep, sweating serves a temperature regulation function, and the sweat glands active during sleep are controlled exclusively by the sympathetic nervous system, your body’s “fight or flight” system.
What makes this useful for sleep tracking is that the skin’s electrical activity produces distinct patterns during different sleep stages. Researchers have identified specific “storms” of electrical activity, periods lasting at least a minute with clusters of measurable skin voltage changes, that increase during the third cycle of REM sleep each night. These storms also appear more frequently and with greater intensity in people with obstructive sleep apnea, because apnea episodes trigger unexpected bursts of sympathetic nervous system activity. Evaluating the length and energy content of these storms turns out to be especially useful for detecting severe sleep apnea.
How Your Sleep Score Is Calculated
After collecting all this raw data, your tracker distills it into something readable, usually a score out of 100. Fitbit, for example, calculates its sleep score from three components: sleep duration (how long you actually slept), sleep quality (how much time you spent restless or awake), and restoration (based on heart rate and sleep stage proportions). Other brands use slightly different formulas, but most weigh the same core factors.
The score is meant as a quick snapshot, not a clinical measurement. Two nights with the same score might feel very different depending on factors the tracker can’t measure, like how much caffeine you had, whether you were stressed, or whether you were fighting off a cold. Still, tracking your score over weeks can reveal patterns: consistently low scores on certain days of the week, or gradual improvements after changing a habit.
How Accurate Are These Estimates?
The gold standard for sleep measurement is polysomnography (PSG), a clinical test that monitors brain waves, eye movements, muscle activity, and more. Consumer trackers can’t measure brain activity, so they’re always working from indirect signals. A 2024 systematic review comparing three popular trackers to PSG found that total sleep time estimates were impressively close in some cases. WHOOP disagreed with the clinical measurement by only 1.4 minutes on average for total sleep time. Fitbit Charge 4 showed the least disagreement for REM sleep, off by about 4 minutes.
Stage-by-stage accuracy is more uneven. Fitbit Charge 4 correctly identified deep sleep about 75% of the time and REM sleep about 86.5% of the time, outperforming the Garmin Vivosmart 4 and WHOOP on those measures. But all three devices had notable gaps. WHOOP overestimated REM sleep by 21 minutes per night on average.
The general pattern across studies is that wearable trackers tend to overestimate total sleep time and how quickly you fell asleep, while underestimating how much time you spent awake during the night. In other words, your tracker is slightly optimistic. It’s more likely to mistake quiet wakefulness for sleep than the reverse.
What Can Throw Off Your Tracker
Several real-world factors reduce accuracy beyond what shows up in controlled studies. Sharing your bed with a partner or pet can introduce movement that confuses the accelerometer. Darker skin tones and tattoos on the wrist can interfere with the optical heart rate sensor, since it depends on light penetrating and reflecting off skin. People with insomnia face a particular irony: the condition itself degrades tracking accuracy, because insomnia involves long periods of motionless wakefulness that trackers are prone to misclassifying as sleep.
Fit matters too. A loose band allows ambient light to leak under the sensor, corrupting heart rate and oxygen readings. Wearing the tracker snug, about a finger’s width above your wrist bone, improves data quality significantly.