Single cell activity was recorded from the putamen in rhesus monkeys (Macaca mulatta) performing limb movements made both with and without a preceding preparatory "set" concerning the impending direction of movement. The monkeys were trained to perform an elbow step-tracking task, each trial of which required the subject to make two sequential movements that were virtually identical in terms of their direction, amplitude, velocity and muscle pattern. The first movement was preceded by a foreperiod, the "pre-instruction" period, during which the monkey could not predict the direction (flexion/extension) of the forthcoming movement. The second movement was also preceded by a variable foreperiod, the "post-instruction" period, but during this interval the monkey was aware of the impending direction of movement because it was always required to match that of the first movement. To dissociate the directions of motor set (and limb movement) from the patterns of tonic (and phasic) muscular activation associated with task performance, some trials included the application of torque loads that either opposed or assisted the movements required by the behavioral paradigm. Two principal forms of task-related activity were observed among the 232 putamen neurons (all located within the region of arm representation) whose discharge patterns showed significant relations to one or more features of the behavioral paradigm. "Movement-related" neurons (178/232, 76.7%) discharged selectively in association with either flexion or extension movements of the elbow. "Set-related" neurons (47/232, 20.3%) manifested sustained, selective changes in discharge throughout the post-instruction period, while the monkey was required to maintain a preparatory set concerning the impending direction of movement. Extensive limb and axial EMG recordings showed no evidence of task-related activity patterns comparable to those that characterized the set-related neurons, and scleral search coil recordings showed no relationship between set-related neuronal activity and either the direction of gaze or the direction and timing of eye movements during task performance. Only a small percentage (7/232, 3.0%) of the task-related neurons showed both movement- and set-related activity. In most cases (37/47, 79%), set-related activity was directionally selective, with the magnitude of discharge reflecting the impending direction of movement.(ABSTRACT TRUNCATED AT 400 WORDS)