The first purpose of this study was to assess the eventual training adaptations in the time to exhaustion at the same severe velocity occurring after severe interval-training programs in few- and well-trained subjects. In the event of such training adaptations, the second purpose was to identify the discriminant factors of performance improvement according to the initial training status. Seven few- and six well-trained subjects performed: firstly, an incremental test to determine the maximal oxygen consumption (VO 2max ), the energy cost of running (ECR), the velocity associated with the achievement of VO 2max ( v VO 2max ) and the lactate threshold (LT expressed in VO 2 , km.h -1 , % v VO 2max ); secondly, an all-out test at the velocity corresponding to the midway between v LT and v VO 2max ( v ?50) to determine the time to exhaustion ( t max ); such tests were carried out before and after 4- and 8-week severe interval-training programs. In the few-trained subjects, all factors of performance (i.e., VO 2max , ECR, v VO 2max , LT expressed in VO 2 , km.h -1 , % v VO 2max ) and t max at the pretraining v ?50 were improved after training (+8, -8, +7, +9, +14, +6% and +79%, respectively); only the increase in v LT was related to the one in t max ( r = 0.714, p = 0.05, n = 7). In the well-trained subjects, only v VO 2max was improved (+3%) due to the decrease in ECR (-3%), t max at the pre-training v ?50 did not vary after training; only the three subjects (over six) who improved their v LT (+0.5, +0.5, +0.8km.h -1 , respectively) improved their t max (+10, +24, +101%, respectively) ( r = 0.895, p = 0.01, n = 6). So, whichever the initial training status, any training-induced adaptation in v LT appeared as a major factor of performance improvement especially at supra-LT velocities.