Smartwatches miss sweat loss by 25 to 33%

The most quotable recent paper on wearable hydration is Carrier et al., "Evaluation of Exertional Sweat Loss Estimates in Wearable Technology," published in Sports Health in 2025.

The team ran 111 participants through up to four running trials each — indoors and outdoors, 2.5 to 20 km, including interval sessions — and compared two commercial smartwatches' sweat-loss estimates against gold-standard body-mass measurement.

The devices were not useless. Concordance correlation ran from 0.71 to 0.90, meaning they follow the general shape of a session. But absolute error was large and lopsided: mean absolute percentage error of roughly 25% for the Samsung device and 33% for the Garmin.

The authors' conclusion is direct. Wearables are "promising and convenient" for general tracking but "lack the precision to replace laboratory methods for hydration management."

A device that can be a third of a litre wrong is useful for spotting a trend. It is not an instrument you would use to decide whether a specific athlete on a specific afternoon needs to come off the field.

Why the ceiling exists

These watches do not measure fluid. They model it from heart rate, motion and ambient temperature. The error is not a calibration bug that a firmware update fixes; it is inherent to inferring a quantity from proxies that correlate with it loosely.

The alternative is to measure a physiological quantity directly, which is what the sweat-sensor research has been working to validate in the field rather than the lab. A 2025 npj Digital Medicine paper (Vol. 8, Art. 76) describes a wearable microfluidic biosensor that directly reads whole-body sweat loss, sweat rate, sodium concentration and sodium loss, alongside skin temperature, thermal flux and motion. It was built and tested for uncontrolled conditions — 40–65°C, 80–100% humidity, one to twelve hours of wear — addressing a real gap: very little sweat-sensor work had been validated against ground truth outside a climate chamber.

What the field data showed

In a Chevron safety study run with UC Berkeley's School of Public Health under NIOSH funding, connected hydration devices were worn by 42 workers across 411 offshore and onshore summer shifts.

Two findings stand out. On-body temperature ran 6–12°F hotter than ambient readings, and most shifts met or exceeded NIOSH's 91°F heat-stress threshold — direct evidence that the standard Heat Index systematically understates individual thermal load. And the hydration alerts changed behaviour: workers drank more when prompted.

That second point is the one worth holding onto. A measurement is only worth the intervention it triggers.

The regulatory backdrop

OSHA's Heat Injury and Illness Prevention rule was proposed on 30 August 2024, moved through public hearings, and closed its post-hearing comment period on 30 October 2025. As of mid-2026 it is not finalized, and employers are advised not to plan around an imminent federal standard.

What did happen: OSHA's original Heat National Emphasis Program was renewed as an expanded program running through April 2031, so enforcement continues without a final rule. The proposed thresholds are concrete — water and shade access plus one quart per worker per hour at an 80°F heat index, and mandatory paid rest breaks at 90°F.