Six male subjects took part in a pilot study on a stationary wheelchair ergometer. They propelled the ergometer at a speed of 0.55, 0.83, 1.11 and 1.39 m/s. The speed increased every 3 min. Inertia and friction force were adjusted proportional to body weight. Every third minute 750 samples of the torque and velocity signals were digitized at a sampling rate of 100 Hz. From the signals mean external power output (Pmean), peak power (Ppeak), mean torque (Mmean) and peak torque (Mpeak), work/cycle, 'time-to-peak torque' (TTP), cycle duration (CT), push time (PT) and recovery time (RT) were determined in relation to mean velocity (speed). For the mean velocity range studied, analysis of variance (P less than 0.05) revealed significant increments in Ppeak, Mpeak, Pmean, Mmean and work/cycle with increasing mean velocity, whereas CT and PT showed a significant decrease. TTP showed a decrease with speed which, however, was not statistically significant. The RT showed no significant variation as well. Our previous research into propulsion techniques mainly focused on movement frequency and timing and was conducted during wheelchair ambulation on a motor driven treadmill. Despite considerable interindividual variation in terms of movement pattern, current and previous studies showed similar trends in the timing pattern (cycle, push, recovery duration) with respect to speed. Theoretical considerations regarding variations in peak torque and work/cycle with respect to velocity are supported by the current results. Both torque and work/cycle are important technique parameters and of relevance in speed regulation. The data also suggest that wheelchair ambulation can be validly simulated and studied with the special purpose wheelchair ergometer.(ABSTRACT TRUNCATED AT 250 WORDS)