During prolonged and intense running exercises beyond the critical power level, a VO2 slow component elevates VO2 above predicted VO2-work rates calculated from exercise performed at intensities below the lactate threshold. In such cases, the actual VO2 value will increase over time until it reaches VO2max. The aims of the present study were to examine whether the VO2 slow component is a major determinant of VO2 over time when running at a speed beyond critical velocity, and whether the exhaustion latency period at such intensity correlates with the magnitude of the VO2 slow component. Fourteen highly trained long-distance runners performed four exhaustive runs, each separated by one week of light training. VO2 and the velocity at VO2max (vVO2max) were determined for each by a graded treadmill exercise. The critical velocity (86.1 +/- 1.5% vVO2max) of each runner was calculated from exhaustive treadmill runs at 90, 100 and 105% of vVO2max. During supra-critical velocity runs at 90% of vVO2max, there was no significant rise in VO2max (20.9 +/- 2.1 ml min-1 kg-1 between the third and last min of tlim 90), such that the runners reached a VO2 steady-state, but did not reach their vVO2max level over time (69.5 +/- 5.0 vs 74.9 +/- 3.0 ml min-1 kg-1). Thus, subjects' time to exhaustion at 90% of vVO2max was not correlated with the VO2max slow component (r = 0.11, P = 0.69), but significantly correlated with the lactate threshold (r = 0.54, P = 0.04) and the critical velocity (% vVO2max; r = 0.65, P = 0.01). In conclusion, the present study demonstrates that for highly trained long-distance runners performing exhaustive, supra-critical velocity runs at 90% of vVO2max, there was not a VO2 slow component tardily completing the rise of VO2. Instead, runners will maintain a VO2 steady-state below VO2max, such that the time to exhaustion at 90% of vVO2max for these runners is positively correlated with the critical velocity expressed as % of vVO2max.