The best running watches, ranked by what their data is worth

How this ranking was built

Five weighted factors, applied identically to every device on this site: evidence (30%), hardware (20%), metric coverage (20%), endurance (15%) and cost per capability (15%). The weights and the reasoning are published at How we rank, and every pick below has an expandable breakdown showing exactly which sub-score produced its number.

We have not worn any of these watches. There is no test lab here and no drawer of review units. What follows comes from published validation research, the manufacturers' own specification pages, and arithmetic you can repeat.

The finding that reorders this list

The intuitive ranking of these six watches is by price, and the intuitive read is that the Forerunner 570 is the sensible-value choice against its more expensive sibling. Divide price by manufacturer-stated GPS endurance and that flips.

Garmin states 26 hours of GPS mode for the 970 and 18 hours for the 47 mm 570. At today's live prices that works out at $28.85 per GPS hour for the 970 against $30.55 for the 570. The more expensive watch is the cheaper one, per hour of the thing you actually bought it to do.

That does not automatically make it the right purchase. If your longest run is ninety minutes, hours 19 through 26 are worth precisely nothing to you and the 570 is the better buy. But it reframes the question from "is the 970 worth $200 more?" to "do I need maps, and do I need the battery?" — which is something you can actually answer about yourself.

And then the COROS PACE 4 walks in and states 41 GPS hoursin All-Systems mode, in a 32-gram case, for about a third of the 970's money. On cost per GPS hour it is not close — it scores a perfect 10 on both endurance and cost, the only device on the site to do so.

It still only places third, and the reason is worth understanding because it is the rubric working as intended. Endurance and cost carry 30% of the score between them; evidence carries 30% on its own. The PACE 4 wins the first pair outright and has the weakest evidence position in the top four, so the Polar Vantage M3 — with a genuinely favorable (if dated) peer-reviewed positioning result behind it — edges past it into second on the weighted total. Different weights would give a different order, which is exactly why ours are published rather than asserted.

Why the top pick is not the one with the best evidence

Here is something worth being blunt about: none of these watches has an independent validation of its optical heart rate.Not one. We searched, and the only measured test of any current-generation running watch sensor we found was the5krunner's comparison of the COROS PACE 3 — a previous model, with a dataset the author himself flags as limited, and an unfavorable result.

So the evidence sub-scores across this roundup are low and closely bunched, and they are low for an honest reason rather than a punitive one. The two Polars score highest on evidence because Polar is the one brand with a favorable peer-reviewed positioning result — and even that is a 2020 study of the Vantage M, Vantage V and V800, which are not the watches sold today. It is brand engineering history, not a measurement of the M3, and we score it accordingly.

The Suunto Race S is the clearest illustration of the rubric doing its job. Excellent hardware, the cheapest dual-frequency AMOLED here, and no published validation of any kind that we could find. It ranks fifth. If you think that is unfair to a good watch, you may well be right — but you can see exactly which number did it, which is more than any other ranking in this category will give you.

If you only remember one thing

If you came here because your heart rate readings are wrong, do not buy any of these watches. The problem is that an optical sensor is reading through a moving wrist joint, and it is not brand-specific: in the 2025 Sensors placement study, a wrist device showed a bias of roughly 33 bpm during explosive work while the same sensor moved up the arm performed far better. A $105 chest strap fixes that with the watch you already own. We would rather tell you that than sell you a watch that will not help — the full argument is on the heart-rate upgrade path.

What to actually check before you buy

Multi-band, or single-band

This is the one specification that genuinely changes what the watch records. A single-band receiver listens on one frequency and cannot easily distinguish a signal that arrived directly from one that bounced off a building. A dual-frequency receiver hears both L1 and L5, and the disagreement between them is the clue that lets it discard the reflection. If your tracks wander in cities or under tree cover, this is the fix. If they do not, it is a specification you are paying for and will not notice.

Which battery number you are reading

Manufacturers publish several endurance figures per watch, and they differ by up to a factor of two. "GPS-only" is the flattering one. "All-systems multi-band" — everything switched on, which is how you would run it if accuracy is why you bought it — is considerably shorter. Every battery figure on this site is labeled with the mode it refers to. Not every spec sheet you read elsewhere will be.

Maps, which are the real difference in the Garmin range

The gap between the 570 and the 970 is maps, titanium and battery. It is not sensors and it is not the data. Both record the same core metrics from the same generation of sensor hardware. If you run known routes, the map is a beautiful feature you will use twice.

Weight, which matters more the longer you go

The PACE 4 is 32 grams with the nylon band; a flagship multisport watch is roughly double that. Over an hour it is irrelevant. Over a hundred miles it is the difference between forgetting you are wearing it and being aware of it every single step. This is one of the few things a spec sheet tells you honestly and directly.