Improvement of exercise capacity by continuous (CT) versus interval training (IT) remains debated. We tested the hypothesis that CT and IT might improve peripheral and/or central adaptations, respectively, by randomly assigning 10 healthy subjects to two periods of 24 trainings sessions over 8 weeks in a cross-over design, separated by 12 weeks of detraining. Maximal oxygen uptake (VO2max), cardiac output (Qmax) and maximal arteriovenous oxygen difference (Da-vO2max) were obtained during an exhaustive incremental test before and after each training period. VO2max and Qmax increased only after IT (from 26.3 +/- 1.6 to 35.2 +/- 3.8 ml min(-1) kg(-1) and from 17.5 +/- 1.3 to 19.5 +/- 1.8 l min(-1), respectively; P < 0.01). Da-vO2max increased after both protocols (from 11.0 +/- 0.8 to 12.7 +/- 1.0; P < 0.01 and from 11.0 +/- 0.8 to 12.1 +/- 1.0 ml 100 ml(-1), P < 0.05 in CT and IT, respectively). At submaximal intensity a significant rightward shift of the Q/Da-vO2 relationship appeared only after CT. These results suggest that in isoenergetic training, central and peripheral adaptations in oxygen transport and utilization are training-modality dependant. IT improves both central and peripheral components of Da-vO2max whereas CT is mainly associated with greater oxygen extraction.