The purpose of this study was to investigate the effect of two cycling velocities on power output and concomitant metabolic and cardiorespiratory responses to repeated all-out exercises. Mean power output (P(m)), total work ( W(tot)), total oxygen consumption (VO(2tot)) and blood lactate accumulation (delta[La](b)) were evaluated in 13 male subjects who performed two series of twelve 5-s bouts of sprint cycling. Recovery periods of 45-s were allowed between trials. One series was executed at optimal velocity (V(opt): velocity for greatest power) and the other one at 50% V(opt) (0.5 V(opt)). Velocities obtained in these conditions were V(opt=)116.6 (4.7) rpm; 0.5V(opt)=60.6 (4.9) rpm. After a phase of adaptation in oxygen uptake in the first part of the series, the data from the 6th to the 12th sprint were as follows: P(m), 924.6 (73.9) versus 689.2 (61.8) W; W(tot), 29.95 (4.14) versus 22.04 (3.17) kJ; VO(2tot), 12.80 (1.36) versus 10.58 (1.37) l; delta[La](b), 2.72 (1.22) versus 0.64 (0.79) mmol x l(-1), respectively (P<0.001). Both W(tot) and VO(2tot) were consistently higher at optimal velocity (+21 and +35.8%, respectively). The present findings demonstrate that during intermittent short-term all-out exercise requiring maximal activation, the energy turnover is not necessarily maximal. It depends on muscle contraction velocity. The increase, lower than expected, in metabolic response from 0.5 V(opt) to V(opt) suggests also that mechanical efficiency is higher at V(opt).