This work compares the specific effects of 3 mo of moderate, isometric, or dynamic voluntary exercises on the contractile properties of human adductor pollicis muscle. Isometric training consisted of 10 daily contractions of 5-s duration at the frequency of one contraction per minute. Dynamic training consisted of 10 daily series of 10 fast contractions (less than 0.5-s duration) moving a load of one-third of the maximal muscle strength at a frequency of one series per minute. Both training programs produced a concomitant increase in maximal tetanic tension and in peak rate of tension development (Ro). A larger increase (P less than 0.05) was found after isometric training (20 vs. 11% after dynamic exercises), whereas Ro augmented more (P less than 0.05) after dynamic contractions (31 vs. 18% after isometric training). Enhancements of twitch force (Pt), rates of twitch tension development (Rt), and of relaxation (St) were, respectively, 20, 20, and 12% after isometric training. There was no modification of contraction time and time of half relaxation (T 1/2R). Conversely, dynamic training produced increases of Rt (25%) and St (16%), associated with an apparently paradoxical decrease of Pt (10%) and reductions of contraction time (11%) and T 1/2R (9%). Maximal shortening velocity was only increased after dynamic training (21%), whereas the maximal muscle power presented a large increase (P less than 0.05) after isometric exercises (51 vs. 19% after dynamic exercises) and a shift of its optimal peak toward heavier loads. This study suggests that human muscle adapts differently to isometric or to dynamic training programs and provides evidence that its contractile kinetics can be altered by exercises performed in physiological conditions.