Seven young, healthy male subjects performed maximal exercise on a cycloergometer with central venous and arterial catheters, before and after autologous retransfusion of red blood cells. Maximal oxygen consumption (VO2max), blood gas composition and haemodynamic variables were measured, in order to test the hypothesis of monofactorial vs. polyfactorial VO2max limitation. Autologous blood retransfusion led to significant increases in haemoglobin concentration and consequently arterial oxygen concentration during maximal exercise, while maximal cardiac output, heart rate and stroke volume were not significantly changed. The relationship between maximal oxygen delivery (cardiac output.arterial oxygen concentration; (Q.CaO2)max and maximal oxygen consumption in this study was VO2max (L.min-1) = 0.02 + 0.64.(Q.CaO2)max (L.min-1), the slope being significantly less than unity. These results suggest that (Q.CaO2)max plays but a fractional role in limiting VO2max, in agreement with recent models concerning the resistance to oxygen flow in the respiratory system (di Prampero and Ferretti, Respir. Physiol. 80: 113-128, 1990). The relative increase in VO2max after blood retransfusion matched the relative increase in 'aerobic performance', measured as the maximal power output that could be maintained aerobically for 30 min. Furthermore, the increase in maximal power output (15 +/- 3 watts) could account for almost all of the extra oxygen consumption. This match suggests that there is an inability to fully utilize muscle oxidative capacity in the normocythaemic state.