Apple Watch VO2 Max: How Cardio Fitness Works and How Accurate It Is
How Apple Watch estimates VO2 max (Cardio Fitness), where to find it in the Health app, how accurate it is against lab tests, how to raise it and how it compares to Garmin.
By Paul Kendrick, Cardio & Endurance Editor · Updated 20 November 2026
The Garmin race predictor takes your VO2 max estimate, checks it against several weeks of your training, and spits out a finish time for the 5K, 10K, half marathon and marathon. It's a decent guide for shorter races. For the marathon it tends to run fast, sometimes by a lot, and the people it flatters most are runners with good aerobic fitness but not much mileage behind them. Below is what's going into those numbers, why the long-distance ones drift, and the handful of fixes that actually make them more believable.
Garmin is fairly guarded about the maths, but its own documentation gives the outline. The Forerunner 265 manual says the watch "uses the VO2 max. estimate and your training history to provide a target race time", and that it "analyzes several weeks of your training data to refine the race time estimates" (Forerunner 265 manual, Viewing Your Predicted Race Times). The same wording turns up in the manuals for the Forerunner 165 and 970, so it's the standard current approach rather than a 265 quirk.
Garmin's technology page adds a bit more. It describes VO2 max as the foundation, combined with real-world data on how long people can sustain different intensities, and says the prediction is refined by looking at trends in your training load and mileage over the last few weeks, with "special consideration" given to the types and frequency of runs you've done (Garmin, Race time prediction).
So there are really two layers:
Older Garmins leaned almost entirely on the first layer; manuals for models like the Forerunner 245 and 945 describe the prediction as coming from VO2 max and "published data sources" alone. The training-history layer is the newer addition, which is why predictions on current watches move around more from week to week than they used to.
If you add a race to your Garmin Connect calendar, newer watches also show a countdown with your goal or predicted finish time for that event, alongside weather and, where course data exists, an elevation profile. That's a separate race-day screen; the four standard distances on the performance glance assume a flat course in normal conditions.
It depends heavily on the distance. A sensible rule of thumb: trust the 5K, sanity-check the 10K, treat the half as a stretch target and the marathon as a best-case scenario.
The reason is physiological rather than a bug. Over 5K you're running close to your VO2 max for 20 to 30 minutes, so an aerobic-capacity estimate maps pretty directly onto race pace. A marathon is a different test: it's run well below VO2 max, and what decides your time is how much of that capacity you can sustain for three or four hours, plus whether your glycogen stores, muscles and pacing survive the last 10K. Two runners with identical VO2 max can finish a marathon 30 minutes apart. The watch sees the same engine in both and has only your training log to tell them apart.
The same problem shows up in the classic pen-and-paper approach. Most race calculators, and a lot of the logic behind watch predictions, descend from the formula Peter Riegel published in 1981 from world-record data: your new time equals your old time multiplied by the distance ratio raised to roughly 1.06. Plug in a 50-minute 10K and it gives you about 1:50:20 for the half and 3:50:00 for the marathon.
A study of 2,303 recreational runners tested that formula against real results (Vickers and Vertosick, BMC Sports Science, Medicine and Rehabilitation, 2016). It held up well for races up to the half marathon, then "dramatically underestimated marathon time", giving times at least 10 minutes too fast for half the runners. Weekly mileage was one of the strongest predictors of how fast people actually ran: in their data, 50 miles a week rather than 30 went with a marathon roughly 25 minutes quicker for men and 32 minutes quicker for women. That's an observational link, not proof, but it lines up with what most coaches would tell you.
Garmin's training-history layer exists to correct for exactly this gap. In practice it narrows it rather than closing it, and the runners left furthest out tend to be those whose fitness has outrun their volume: people who race parkrun hard every Saturday but rarely run longer than 10 miles, or anyone who's come into running from another endurance sport.
| Distance | How far to trust it | Why |
|---|---|---|
| 5K | Usually close if your VO2 max reading is sound | Raced near VO2 max, so the estimate maps directly |
| 10K | Reasonable, may be a little fast | Still a mostly aerobic-capacity effort |
| Half marathon | Often optimistic for lower-mileage runners | Endurance and pacing start to matter more |
| Marathon | Frequently too fast unless you're doing proper long runs | Fuelling, leg durability and volume decide the result |
The predictor can also err the other way. Heat, hilly routes and a lot of slow, easy running can all drag your VO2 max down, and with it every predicted time. If you've just set a 10K PB that beats the watch's prediction, believe the race.
Before blaming the algorithm, check the inputs. Most wildly off predictions come from one of these:
None of these turn the watch into a crystal ball, but together they remove most of the obvious errors.
Set your max heart rate from real data. The age-based default (220 minus age, or similar) can be out by 10 to 15 beats either way. Use the highest figure you've seen in a hard race finish or a structured test, then enter it manually. Our heart rate zone training guide walks through how to find it safely.
Wear a chest strap for key runs. A strap reads the heart's electrical signal rather than shining light through your wrist, so it doesn't get confused by arm swing or cadence. You don't need it every day; wearing one for tempo runs, intervals and races gives the watch its cleanest data where it matters most. The Polar H10 is our pick and pairs with any recent Garmin.
Check Polar H10 price on AmazonRun long, regularly. If you want a believable marathon prediction, the watch needs to see marathon-style training: a weekly long run that builds towards 18 to 20 miles, and enough total mileage to support it. This isn't gaming the algorithm; it's the same training that makes you able to run the time.
Record runs properly. Use the outdoor Run profile with a GPS fix for most sessions, and complete your user profile (age, sex, weight). Garmin's manuals single out the profile and max heart rate as the two things to get right for the most accurate estimate.
Give it time. Predictions settle after a few weeks of consistent running. Judging them after your first three runs on a new watch isn't fair on the algorithm or on you.
Use it as a check, not a pacing plan
A good way to use the predictor is as a cross-check against a recent race. If your watch says 3:30 but your last half marathon was 1:48, a realistic marathon target sits nearer 3:50 to 4:00 for most club-level runners. Pace from the race result, then watch whether the prediction drifts towards it as your training builds.
Garmin doesn't publish one master list, so the safest answer comes from each model's own manual or spec sheet. Ones we've confirmed:
Older Forerunners (245, 945 and others) and fenix and epix models also show predicted race times, though older firmware may use the simpler VO2 max-only method. Lifestyle watches such as the Venu 3S don't include it. If you're unsure about yours, search your model's manual for "Viewing Your Predicted Race Times".
Check Forerunner 265 price on AmazonIf you're choosing a watch partly for training features like this, our best Garmin watch UK guide sorts the range by budget, and the Garmin Body Battery guide explains the recovery side of the same data.
It starts from your VO2 max estimate, then adjusts using your training history. Garmin's manuals say the watch analyses several weeks of training data to refine the times, and Garmin's own technology page says it looks at trends in training load and mileage, including the types and frequency of your runs.
For 5K and 10K it's usually in the right area if your VO2 max reading is sound. For the marathon it's often too fast, especially for runners without much weekly mileage or many long runs, because aerobic fitness alone doesn't tell the watch whether your legs and fuelling will hold up for 26.2 miles.
Most likely because your VO2 max is decent but your endurance base isn't. A good 10K runner on 20 miles a week can't hold the same relative pace for a marathon as someone running 50. A max heart rate set too high, or a wrist sensor reading low, also inflates VO2 max and drags every prediction faster.
The predictor needs a VO2 max estimate first, and that only comes from outdoor runs with GPS and heart rate, at a reasonable effort, for at least 10 minutes. Garmin also warns the predictions may look wrong at first because the watch needs a few runs to learn your performance.
Garmin's manuals list predicted race times on current Forerunners including the 165, 170, 265, 570 and 970, and the Forerunner 55's spec page lists it too. The vivoactive 6 has it; the vivoactive 5 doesn't. Garmin doesn't publish one master list, so check your own model's manual for a Predicted Race Times page.
How Apple Watch estimates VO2 max (Cardio Fitness), where to find it in the Health app, how accurate it is against lab tests, how to raise it and how it compares to Garmin.
Fitbit Charge 6 vs Inspire 3 compared: built-in GPS, ECG, Google Wallet, battery, size and price. Our clear verdict on which tracker suits you in the UK.
Fitbit Charge 6 vs Versa 4 compared on screen, ECG, heart rate broadcast, calls, battery, GPS and price, with a clear verdict on which one UK buyers should pick.
Fitbit Versa 4 vs Sense 2 compared: same design, GPS and 6+ day battery, but the Sense 2 adds ECG and a cEDA stress sensor. Which is worth the extra £45?
Best Exercise is a participant in the Amazon EU Associates Programme, an affiliate advertising programme designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.co.uk. As an Amazon Associate we earn from qualifying purchases. This comes at no extra cost to you and never influences our independent reviews or rankings.