Abstract Many studies have found differences in movements made to either side of the body midline. A popular interpretation of these differences has been that movements made by the arm, which is on same side of space in which the visual target appeared, are faster and better organised because they are processed within-hemisphere. Carey et al. (Experimental Brain Research 112 (1996) 496) showed that hemispatial movement differences cannot be accounted for by such a model. Their data suggested that biomechanical factors such as those proposed by Gordon et al. (Experimental Brain Research 99 (1994) 112) could better account for differences in movement duration and several characteristics of velocity and acceleration. The present study examines these arguments by requiring subjects to make rapid pointing movements in two experiments. In the first, results demonstrated that hemispatial effects occurred in pointing movements made without any visual target or vision of the limb. These findings suggest that intra- and inter-hemispheric models are untenable. Gordon et al. argued that hand path direction relative to the long axis of the upper arm accounts for hemispatial effects on kinematics. In the second experiment hand path direction and hemispace were dissociated. Contralateral movements were performed more efficiently than ipsilateral movements, when target and starting positions required an adductive movement to acquire the contralateral target and an abductive movement to acquire the ipsilateral target. These results provide strong support for the Gordon et al. model, although the possible contributions of other dynamic factors and/or differential control of proximal and distal muscles by the central nervous system cannot be ruled out.