1. Regional cerebral blood flow (rCBF) was measured using positron emission tomography in six normal volunteers while at rest and while performing four different repetitive movements of the right arm. 2. The four movements were performed in random order and consisted of abduction of the index finger, making a fist, sequential thumb to digit opposition, and shoulder flexion. All the movements were done at the same rate, using an auditory cue and involved displacements through similar amounts of the physiological range at each joint. 3. Increases in rCBF were interpreted as evidence of local neural activation and all four movements were associated with significant increases in CBF in the contralateral sensorimotor and premotor areas and in the supplementary motor area (SMA). 4. The average increase in blood flow in the contralateral sensorimotor cortex was significantly greater for the shoulder movement (31%) than for the three other movements. The increases with finger opposition (21%) and fist-making (24%) were not significantly different, and both were significantly greater than with index finger movement (13%). These data indicate that neither "fractionation" nor distal movement per se cause selective activation of sensorimotor cortex. 5. Significantly greater increases in blood flow in both the contralateral premotor cortex and the SMA ("nonprimary motor areas") occurred with shoulder movement than with the other movements. Because this difference may be related to the significantly greater activation occurring concurrently in the sensorimotor cortex, this finding does not prove unequivocally a "selective" role of the nonprimary motor areas in proximal movement. 6. Neither of the two nonprimary motor areas showed selective activation when a simple sequence of finger movements was performed compared with repetitive contractions of the same fingers. 7. Shoulder movement alone was associated with significant increases in rCBF in the ipsilateral sensorimotor cortex (10%), the superior vermis of the cerebellum (19%), and Brodmann areas 5 and 40 in the contralateral hemisphere. 8. The average location of the center of excitation in the sensorimotor cortex and SMA differed for the four movements and was interpreted as evidence of within-limb somatotopy. The shoulder focus lay highest in the sensorimotor cortex and lowest in the SMA.