This study presents the concept of constant (iso-) accelerative and decelerative exercise and compares concentric and eccentric torque output during isoaccelerative and isodecelerative movements with that during comparable constant velocity (isokinetic) conditions. Twelve men (19-42 yr) performed maximal voluntary concentric and eccentric knee extensions at velocities of 120 and 240 degrees.s-1 (isokinetic) and at accelerations of 180 and 720 degrees.s-2 (both isoaccelerative and isodecelerative) between 10 degrees and 90 degrees knee angles. At 50 degrees, the 180 and 720 degrees.s-2 tests had velocities of 120 and 240 degrees.s-1, respectively, and thus torque comparisons could be made at a corresponding position and velocity. No difference was seen among the isoaccelerative, isodecelerative, or isokinetic angle- and velocity-specific torques for either the concentric or eccentric tests (P greater than 0.05). The results demonstrated that, under conditions of maximal voluntary effort, movement speed as such (within the range studied) was the essential determinant of muscle force--not whether this speed was attained during accelerative, decelerative, or constant velocity movements. As a testing and training modality, the controlled acceleration technique, particularly eccentric deceleration and concentric accleration, appears to offer advantages as compared with existing methods, since it more faithfully reflects the contraction conditions during natural strength-requiring movements.