Abstract : The purpose of this study was to assess the reproducibility and estimation of maximal (MVC) and submaximal muscular torque. Eight volunteers performed knee extension movements on a dynamometer. After one practice session, 3 test sessions were performed at 1 and 7 day intervals. At each session, subjects performed maximally and submaximally (25, 50 and 75% during isometric (45°) and concentric actions at 90 and 300°/s. The maximal muscular torques revealed no significant difference over the 3 test sessions. When estimating submaximal muscular torques, subjects correctly estimated torque percentages produced in the isometric action (except in the 75% MVC) and in dynamic action (except in the 90°/s, 25% MVC). As for the stability of these estimations, subjects corrected the error in concentric actions evidencing the possible involvement of efferent copy that could enable a comparison between actual and perceived performance. On the other hand, the error in isometric actions was maintained pointing out to very limited use of this contraction mode in sporting activities.