Prehension movements of the right hand were recorded in normal subjects using a computerized motion analyzer. The kinematics and the spatial paths of markers placed at the wrist and at the tips of the index finger and thumb were measured. Cylindrical objects of different diameters (3, 6, 9 cm) were used as targets. They were placed at six different positions in the workspace along a circle centered on subject's head axis. The positions were spaced by 10 degrees starting from 10 degrees on the left of the sagittal axis, up to 40 degrees on the right. Both the transport and the grasp components of prehension were influenced by the distance between the resting hand position and the object position. Movement time, time to peak velocity of the wrist and time to maximum grip aperture varied as a function of distance from the object, irrespective of its size. The variability of the spatial paths of wrist and fingers sharply decreased during the phase of the movement prior to contact with the object. This indicates that the final position of the thumb and the index finger is a controlled parameter of visuomotor transformation during prehension. The orientation of the opposition axis (defined as the line connecting the tips of the thumb and the index finger at the end of the movement) was measured. Several different frames of reference were used. When an object-centered frame was used, the orientation of the opposition axis was found to change by about 10 degrees from one object position to the next. By contrast, when a body-centered frame was used (with the head or the forearm as a reference), this orientation was found to remain relatively invariant for different object positions and sizes. The degree of wrist flexion was little affected by the position of the object. This result, together with the invariant orientation of the opposition axis, shows that prehension movements aimed at cylindrical objects are organized so as to minimize changes in posture of the lower arm.