What a heart rate zone actually is

The convention, published by the people who sell it

There is no natural law that divides effort into five parts. The five-zone model is a convention, and Polar publishes it plainly in its own training guide as percentages of maximum heart rate.

Notice what that table is: pure arithmetic on a single number. Get that number wrong and every boundary moves together, in the same direction, by the same proportion. Nothing else in the model can compensate.

The number underneath everything, and where it comes from

Polar's own guide says the "most simple way" to establish your maximum heart rate is "the rule 220-your age". That is a population average, not a measurement of you, and it is the default almost everybody ends up training against.

Garmin is more explicit about the mechanics. Its manual states that the watch "uses your user profile information from the initial setup to determine your default heart rate zones", and instructs you that "for the most accurate calorie data during your activity, set your maximum heart rate". It also offers two other anchors: %Max. HR, and %HRR, which the manual defines as maximum heart rate minus resting heart rate. Separately it lets you enter a lactate threshold heart rate, and it can auto-detect that during an activity — but it does not auto-detect your maximum.

Read that sequence carefully. The one input the whole model depends on is the one the watch will not measure for you.

Two errors, stacked

So a zone boundary carries two independent sources of error, and it is worth separating them because they have very different fixes.

• The anchor error. Your true maximum heart rate may be well away from 220 minus your age. If it is 10 bpm out, every boundary is out by roughly 10% of 10 bpm at each threshold — and the whole ladder shifts.
• The reading error. Whatever your zones are, the watch has to tell you which one you are in right now. This is the part that has actually been measured.

The reading error, quantified

The 2025 anatomical-placement study put 28 participants through rest, a cycling warm-up, burpees and a modified Bruce treadmill protocol against a Polar H10 chest strap reference. Two of its findings decide how much a zone readout is worth.

• A Whoop 4.0 worn at the wrist showed limits of agreement of roughly plus or minus 10 to 19 bpm during warm-up.
• A Garmin Forerunner 55 at the wrist recorded a bias of approximately 33 bpm during burpees — the largest error in the study.

Now do the arithmetic against the zone table. Each zone spans 10% of your maximum heart rate. For a maximum of 190 bpm, that is a 19 bpm band. A wrist reading with limits of agreement of 10 to 19 bpm is, at its worst, a full zone wide — during warm-up, which is precisely the part of the session where people check whether they are in zone 2.

Move the sensor and the problem largely disappears. In the same study a Polar Verity Sense on the forearm recorded a bias under 0.1 bpm on the graded treadmill protocol with a concordance of 0.997, and roughly 1.6% mean error at rest. The full position-by-position breakdown is in chest strap versus armband.

Why zone 2 is the most fragile prescription in training

Zone 2 is 60 to 70% of maximum, and it is the zone most modern endurance plans are built around. It is also the one where your body gives you the least feedback: nothing hurts, your breathing is conversational across a wide range, and you cannot tell 68% from 74% by feel. So the number is doing all the work, and an error that would be obvious in zone 5 passes unnoticed here.

That is the honest answer to "which watch is best for zone 2 training". Almost any of them will draw the chart. What decides whether the chart is true is the sensor feeding it and the maximum you entered — neither of which is a watch feature.

Anchoring on threshold instead does not escape the problem

The sophisticated alternative is to set zones from lactate threshold heart rate rather than maximum, which is what Garmin's LTHR field and most coaching software do. It is a better physiological anchor. It is not a more certain one when the watch supplies it.

In the 2025 Frontiers in Physiologystudy of watch-derived threshold estimates against a treadmill graded exercise test with blood lactate sampling, mean absolute error in threshold heart rate was 8.93 bpm for the COROS PACE 3, 10.66 bpm for a Huawei GT Runner and 11.44 bpm for the Garmin Forerunner 265. Correlation with the laboratory value ranged from r = 0.13 to r = 0.67, and no device met the authors' equivalence criteria. We take that apart in what a lactate threshold estimate is worth.

Nine to eleven beats is, again, roughly half a zone. Swapping anchors swaps one estimate for another.

What to actually do about it

Fix the reading first, because it is cheap and it has evidence behind it. A chest strap or a forearm armband costs a fraction of a watch and is the only part of this chain with peer-reviewed support. Nothing else you can buy improves the number as much per dollar.

Then fix the anchor with something you did, not something you are. The highest heart rate you have genuinely seen in a hard race or a maximal hill effort is a better estimate of your maximum than 220 minus your age, and it costs nothing. If your watch-derived threshold and your race results disagree, believe the race.

Then stop treating boundaries as walls. Zone 2 at 68% and zone 3 at 71% are not different physiological states; they are two sides of an arbitrary line drawn on an estimate. The full device-by-device record of what has been measured is in heart rate accuracy by device, and the most accurate sensors we cover are ranked in the most accurate heart rate monitors.

One last framing worth keeping: a fitness tracker and a training watch draw the same zone chart from very different data quality, and the chart looks identical either way. We set out what actually differs in fitness tracker versus smartwatch.

If you train by zones, in priority order

• Get an electrical chest strap or a forearm armband. This is the only change on the list with peer-reviewed evidence behind it, and it is the cheapest.
• Replace 220 minus age with your own highest observed effort. A real number from a real race beats a population average.
• Pick one anchor and keep it. %Max. HR and %HRR produce different boundaries from the same body. Switching between them mid-block makes your history meaningless.
• Re-check once a season, not once a week. Maximum heart rate moves slowly. A zone chart that changes every month is measuring noise.

The three sensors above are ranked on the same site-wide rubric as everything else, with the weights published at how we rank. They are here because fixing the reading is the concrete, purchasable half of this problem — the anchor is free to fix and nobody can sell it to you.