The purposes of this study were (1) to determine the time sustained above 90% of VO2max in different intermittent running sessions having the same overall time run at the velocity (vVO2max) associated with VO2max, and (2) to test whether the use of a fixed-fraction (50%) of the time to exhaustion at vVO2max (Tlim) leads to longer time spent at a high percentage of VO2max. Subjects were 8 triathletes who, after determination of their track vVO2max and Tlim, performed three intermittent running sessions alternating the velocity between 100% and 50% of vVO2max, termed 30 s-30 s, 60 s-30 s, and 1/2 Tlim, where the overall time at vVO2max was similar (= 3 x Tlim). VO2max achieved in the incremental test was 71.1 +/- 3.9 ml.min-1.kg-1 and Tlim was 236 +/- 49 s. VO2peak and peak heart rate were lower in 30 s-30 s than in the other intermittent runs. The time spent above 90% of VO2max was significantly (p < 0.001) longer either in 60 s-30 s (531 +/- 187 s) or in 1/2 Tlim-1/2 Tlim (487 +/- 176 s) than in 30 s-30 s (149 +/- 33 s). Tlim was negatively correlated with the time (in % of Tlim) spent above 90% of VO2max in 30 s-30 s (r = -0.75, p < 0.05). Tlim was also correlated with the difference of time spent over 90% of VO2max between 60 s-30 s and 30 s-30 s (r = 0.77, p < 0.05), or between 1/2 Tlim-1/2 Tlim and 30 s-30 s (r = 0.97, p < 0.001). The results confirm that vVO2max and Tlim are useful for setting interval-training sessions. However, the use of an individualized fixed-fraction of Tlim did not lead to longer time spent at a high percentage of VO2max compared to when using a fixed work-interval duration.