Purpose: The accuracy of heart rate measured with a wrist photoplethysmography monitor can be influenced by the tightening of the wristband, movement of arms, or kinetics of the signal (eg, steady-state exercise vs on- and off-transients). To test these hypotheses, photoplethysmographic and electrocardiographic (ECG) signals were compared. Methods: Thirty participants (50% female) randomly performed two 13′ sequences (3′ rest, 5′ submaximal-intensity exercise, and 5′ passive recovery) on a motorized treadmill and a bicycle ergometer. Heart rate was measured concomitantly with a 10-lead ECG, a chest-strap monitor, and 2 wrist photoplethysmography monitors (Polar Unite) with different tightening (free vs imposed at the maximum tolerable). Results: The level of association (r) and coefficient of variation (CV; ie, the error of measurement) of the Polar Unite versus the 10-lead ECG is affected by the tightness of the wristband (normal vs high; r = .83 and .96, CV = 16.1 and 8.1% for the treadmill, respectively; r = .71 and .97, CV = 20.3% and 6.2% for the bicycle, respectively) by the phase of the signal (transition vs steady state; r = .90 and .97, CV = 9.0% and 7.6% for the treadmill, respectively; r = .93 and .99, CV = 7.5% and 3.1% for the bicycle, respectively) and movement of arms (treadmill vs bicycle; r = .90 and .93, CV = 9.0% and 7.5% during the transition phase, respectively; r = .97 and .99, CV = 7.6% and 3.1% during the steady-state phase, respectively). Conclusion: The accuracy of heart rate measured with a wrist photoplethysmography monitor is affected by the tightness of the wristband and the phase of the signal. A high tightening is required when high accuracy is expected.