Imagine your watch says your threshold pace is 4:16 per kilometer, but the pace you can really hold is closer to 5:00. In a 2025 study, the Garmin Forerunner 265’s guided test read about 2 km/h too fast on average.
That’s a big miss, and it applies to one model and one test. Here’s what the study found, what it didn’t, and how to set running zones you can actually hold.
What the Study Tested on Garmin Lactate Threshold Pace
Researchers led by Lu and colleagues compared three watches to a lab test. The study ran in Frontiers in Physiology in 2025.
Here are the facts that matter:
- The Garmin was the Forerunner 265, run with its guided lactate threshold test and a chest strap.
- Only 23 runners used the Garmin. All 100 runners in the study tested a Huawei watch. Garmin was a smaller add-on group.
- Just 15 of those 23 tests (65%) produced a valid result.
- The lab reference was a treadmill test with fingertip lactate samples and a modified Dmax calculation.
In short: the headline number comes from 15 valid tests on one watch model. It isn’t a verdict on every Garmin.
The study did not test auto-detection from your normal runs. It tested the guided test, where you run at target heart rates for about 25 to 30 minutes. Garmin’s own manual describes that test. The watch blends a fitness estimate, GPS speed and chest-strap heart rate. The exact math isn’t public.
The Numbers: Heart Rate Versus Pace
MAPE is the average miss as a percent of the true value. For Garmin lactate threshold pace, the MAPE was 25.78%. The average miss was 2.17 km/h.
The error was not random. The watch overestimated pace, with a bias of +2.06 km/h. In short: the watch said you could run faster than you really could.
Heart rate looked better. Threshold heart rate had a MAPE of 7.15%, about 11 bpm. The bias was +6 bpm. But the limits of agreement ran from -16 to +29 bpm. The study also did not show that the watch matched the lab.
I won’t call that accurate. It’s closer than pace, and that’s all.
Pace did have one bright spot. Garmin’s pace tracked the lab ranking reasonably well (r = 0.73). Faster runners got faster readings. The scale was just shifted.
Think of a bathroom scale that always reads 8 pounds heavy. It still shows who gained and who lost. It just can’t tell you your real weight.
Why the Sign Changes Between Models
A 2024 study of the Garmin Fenix 7 found the opposite. It used 26 runners and a field-based reference. Its pace came in 11.96% slower than the reference. Heart rate was 1.71% lower, which was not significant.
| Study | Watch | Runners | Pace result | Heart rate result |
|---|---|---|---|---|
| Lu et al., 2025 | Forerunner 265 (guided test) | 23 (15 valid) | 25.78% MAPE, reads fast | 7.15% MAPE, +6 bpm |
| Fenix 7 study, 2024 | Fenix 7 | 26 | 11.96% slower than reference | 1.71% lower, not significant |
Different watch, different test, different reference method, different runners. The error flipped sign.
That’s the real lesson. You can’t fix your watch by adding or subtracting a fixed amount. You have to check it against your own data.
Your watch isn’t lying on purpose. It just can’t promise it’s right.
What a 2 km/h Error Does to Your Intervals
Say your true threshold pace is 5:00 per km, or 12 km/h. A +2.06 km/h bias puts the watch at 14.06 km/h, about 4:16 per km.
That pace is not a threshold. It’s a run you’d fade from in minutes.
| True threshold pace | Watch pace at +2.06 km/h (illustrative) |
|---|---|
| 4:30 per km | about 3:54 per km |
| 5:00 per km | about 4:16 per km |
| 5:30 per km | about 4:38 per km |
| 6:00 per km | about 4:58 per km |
These numbers are illustrative. They apply the average group bias to one pace. Your own miss could be smaller or larger, and the study’s limits ran from -0.21 to +4.33 km/h.
Here’s a fictional but typical case. Dana is 44 and training for a half marathon. Her watch says threshold pace is 4:45 per km. She plans four 8-minute reps there.
Rep one feels hard but fine. Rep two hurts. By rep three she has slowed 10%, and her heart rate is pinned. She finishes the session wrecked and skips her easy run the next day.
Her real threshold was closer to 5:10. She was doing a 5K effort, not threshold work. Hard days that are too hard steal from the aerobic base you build on easy days.
Better Ways to Set Threshold Pace
You don’t need a lab. You need a test you can repeat.
The 30-minute time trial. Run 30 minutes as hard as you can hold, alone, on flat ground. In a study of 27 runners and triathletes, this method predicted lab threshold speed within about 0.76 km/h. Your average pace is your threshold pace estimate. The average heart rate of the last 20 minutes is your threshold heart rate.
Note the different reference methods. That study used a 4 mmol/L lactate criterion, not Dmax. Still, an error under 1 km/h beats the Forerunner’s 2.17 km/h average miss.
Pace at heart rate. Do a steady hard run at your threshold heart rate. Note the pace you held. Repeat it every few weeks. If your pace at that heart rate is dropping, you’re getting fitter. If the watch’s pace is far faster than anything you’ve held there, trust your runs. This is the same idea behind the drift ratio.
A quick sanity check. If you slow more than 5 to 10% by rep three of your threshold intervals, your pace is too fast. Slow it down.
Use your chest strap too. Wrist sensors drift at high effort, which muddies every test. Chest strap versus wrist heart rate matters here more than anywhere.
How AthleteOS Uses This
AthleteOS doesn’t pick one number and call it truth. When you connect your Garmin through workout import, the session analysis compares the watch’s threshold estimate to your own history. It looks at the pace you really held at a given heart rate on hard, steady runs.
If the watch implies a speed you’ve never sustained at that heart rate, AthleteOS flags it. The AI coach then builds your training plan zones from observed efforts. That replaces a single biased estimate with your own evidence.
It doesn’t remove error. Heart rate drifts, weather changes, and a bad night’s sleep moves the number. But it gives you a number you’ve actually run. The same logic applies to Garmin’s VO2max estimate, which has its own limits.
Watches estimate. Your legs measure.
Connect your Garmin at AthleteOS and see how your zones compare to the runs you’ve really done.