Pulse wave velocity measurement is used as an index of arterial stiffness. The purpose was to evaluate the reproducibility of pulse wave velocity measurement at rest, during exercise and recovery from exercise, using an automated device. Twelve healthy young adults (mean age 22.0 +/- 3.1 yrs) underwent an upright submaximal cycle test on two separate occasions, one week apart. Pulse wave velocity, systolic and diastolic blood pressures and heart rate were assessed at rest, during the last 2 min of exercise and 10 min later. Pulse wave velocity was measured on the upper limb and the forearm by the cross-correlation function of photoplethysmography and Doppler signals. Brachial artery pulse wave velocity was calculated from upper limb and forearm pulse wave velocities. No significant difference was found on duplicate measurements of heart rate, systolic and diastolic blood pressures at rest, during exercise and recovery, showing that pulse wave velocity was measured under similar conditions. Coefficient of variation for upper limb and forearm pulse wave velocities ranged from 2.9 to 5.9% at rest and during recovery, and were respectively 2.9% and 8.3% during exercise. However, coefficient of variation for brachial pulse wave velocity was 7.7 and 10.3% at rest, 15.7% during exercise, and 5.8% during recovery. During exercise, pulse wave velocity measurements were satisfying, but indirect assessment of brachial artery pulse wave velocity showed poor reproducibility. Thus, upper limb and forearm pulse wave velocities may be used during exercise to assess the effect of training or drugs on arterial wall mechanical properties.