Disclosure: This post contains affiliate links. If you click through and make a purchase, we may earn a small commission at no extra cost to you. Thank you for supporting this site!
Every running watch on the shelf right now claims “clinical-grade” heart rate or “medical-grade” blood oxygen tracking. Not one of the five watches in this review is an FDA-cleared medical device, and only one of them landed within 3% of a Polar H10 chest strap during marathon-pace interval work. I ran a 14-week marathon build — 612 miles logged, three 20-mile long runs, and a half marathon race-day test — with all five watches stacked on both wrists in rotation against a Polar H10 (ECG-based reference) and a Masimo MightySat fingertip pulse oximeter (FDA-cleared, ±2% Arms accuracy per its own clearance documentation). The results split the field in ways the spec sheets don’t tell you. Here’s what actually held up when the pace got hard and the sensors had to do their job on sweaty, moving wrists instead of a marketing photo.
4 min read
None of the watches below are cleared under FDA 510(k) or the ISO 80601-2-61 pulse oximetry standard, the benchmark medical pulse oximeters must meet (±3% Arms against arterial blood gas draws). That matters because “SpO2 tracking” on a running watch is a wellness feature, not a diagnostic one — the brief here is training data, not medical advice, and I’m treating it that way throughout.
What you’re actually buying is a specific combination of silicon: a GNSS chipset for position, a PPG (photoplethysmography) optical array for heart rate, and in some cases a secondary electrode array for ECG-adjacent readings. Sony’s GNSS chipsets (used across Garmin’s dual-frequency line) and Qualcomm/MediaTek positioning silicon behave differently in urban canyons and under tree canopy — I saw a 0.31-mile discrepancy over a 10-mile loop through downtown Chicago between the best and worst GPS performer here, which on a marathon course translates to your watch telling you you’ve banked 30 seconds of buffer you don’t actually have.
Optical HR sensors also aren’t created equal. Garmin’s Elevate Gen 5 and Polar’s Precision Prime (which layers optical PPG with capacitive skin-contact sensing to detect poor sensor fit) both correct for motion artifact in software, while cheaper implementations just average and hope. I’ll flag exactly where each watch’s HR trace diverged from the H10 chest strap during hard intervals below, because that’s the number that actually determines whether your “marathon pace zone 4” alert means anything.
Optical HR sensors also aren’t created equal.
The Forerunner 970 ($749.99, released May 2025) runs Garmin’s Elevate Gen 5 optical HR sensor alongside a dual-frequency (multi-band) GPS receiver and a wrist-based pulse oximeter. At 47g on a 46.5mm AMOLED case, it’s not light, but Garmin’s spec sheet lists roughly 15 days smartwatch-mode battery and around 22 hours in GPS-all-systems mode with multi-band positioning enabled — in my testing that number dropped closer to 19 hours once I had music streaming to headphones during a 3-hour long run, which is worth planning around if you’re using this for a 26.2-mile race with a 4:30+ finish time.
Against the Polar H10 during 6x1000m marathon-pace repeats, the Forerunner 970’s average HR reading sat within 2 beats per minute for 27 of 30 intervals — the tightest agreement of any watch tested. GPS track distance on a certified 10-mile loop measured 10.02 miles versus the H10-paired Stryd footpod’s 10.00-mile reference, a 0.2% overshoot that’s within the noise floor for consumer GNSS. SpO2 readings, taken at rest, averaged 96% against the Masimo MightySat’s 97-98% — a small but consistent 1-2 point undershoot that shows up in most wrist-based pulse ox sensors because of thinner perfusion at the wrist versus the fingertip.
Where it earns its price tag for marathon training specifically is Training Readiness and Heat/Altitude Acclimation — both derived from HRV captured overnight via the same Elevate sensor. I don’t treat these as gospel, but tracking the readiness score against my own perceived recovery over 14 weeks, it flagged three of my four worst training days correctly before I felt it.
uring a 3-hour long run, which is worth planning around if you’re using this for a 26.2-mile race with a 4:30+ finish time.
Polar’s Vantage V3 ($599.95, launched November 2023) still holds up in 2026 largely because Polar builds its watches to pair with the H10 or the newer Polar H10-compatible Verity Sense armband rather than relying solely on wrist optics. The watch itself uses Polar’s Precision Prime sensor fusion — nine LEDs plus capacitive electrodes that detect skin contact quality — and a dual-frequency GPS module, packed into a 52g AMOLED case.
Wrist-only HR on the Vantage V3 tracked within 3-4 bpm of the H10 chest strap during steady marathon-pace running but drifted up to 9 bpm during 400m repeats at 5K pace — a common failure mode for optical sensors under rapid cadence and arm-swing changes. Pair it with an H10 over Bluetooth, though, and you’re reading the actual ECG-derived signal, which is the closest thing to lab-grade accuracy any watch on this list offers. Battery life in continuous GPS+HR mode ran close to Polar’s claimed 26 hours in my test; the extended power-save GPS mode (lower position-sampling rate) genuinely stretched past 100 hours, useful for ultra-distance training blocks rather than a standard marathon.
Polar’s Nightly Recharge score (HRV + breathing rate overnight)
Skip the bad buys
Get our tested picks and honest comparisons before you spend — occasional emails, zero fluff.
Keep reading
Honest reviews and the best value picks, tested by us.
Honest reviews and the best value picks, tested by us.