Past studies of the kinematics of human prehension have shown that varying object size affects the maximum opening of the hand, while varying object distance affects the kinematic profile of the reaching limb. These data contributed to the formulation of a theory that the reaching and grasping components of human prehension reflect the output of two independent, though temporally coupled, motor programs (Jeannerod 1984). In the first experiment of the present study, subjects were required to reach out and grasp objects, with or without on-line, visual feedback. Object size and distance were covaried in a within-subjects design, and it was found that both grip formation and reach kinematics were affected by the manipulation of either variable. These data suggest that the control mechanisms underlying transport of the limb and grip formation are affected by similar task constraints. It was also observed that when visual feedback was unavailable after movement onset subjects showed an exaggerated opening of their hands, although grip size continued to be scaled for object size. The question remained as to whether the larger opening of the hand during no-feedback trials reflected the lack of opportunity to fine-tune the opening of the hand on-line, or the adoption of a strategy designed to increase tolerance for initial programming errors. To address this question, a second experiment was carried out in which we manipulated the predictability of visual feedback by presenting feedback and no-feedback trials in a random order. In contrast to the situation in which feedback and no-feedback trials were presented in separate blocks of trials (Exp. 1), in the randomly-ordered series of trials presented in Exp. 2, subjects always behaved as if they were reaching without vision, even on trials where visual feedback was continuously available. These findings suggest that subjects adopt different strategies on the basis of the predictability of visual feedback, although there is nothing to suggest that this takes place at a conscious, or voluntary, level. The results of both experiments are consistent with the notion of a hierarchically-organized motor control center, responsible for optimizing performance under a variety of conditions through the coordination of different effector systems and the anticipation of operating constraints.