1. The responses of the first dorsal interosseus (1DI), opponens pollicis (OP), extensor digitorum communis (EDC), brachioradialis (BR), biceps brachii (BB) and anterior deltoid (AD) muscles to magnetic stimulation of the motor cortex were recorded during different motor tasks. 2. Two precision and two power isometric tasks were investigated. The precision tasks were a pincer grip ('grip') and a thrust against a target with the wrist ('push'). In the former, the prime movers were the intrinsic hand muscles, while the proximal muscles played a postural role. In the latter, the prime movers were the proximal muscles. In both tasks, force was controlled through visual feedback. The power tasks required encirclement of a cylinder with the fingers ('grasp'), or sustaining a weight suspended at wrist level ('load'). 3. Magnetic stimulation was applied in eight subjects by a coil placed over the vertex at 1.1-1.2 times the motor threshold for the most excitable muscles. This produced in the prime mover muscles larger motor-evoked responses (MEPs) during grip or push tasks than grasp or load tasks, in spite of similar background EMG levels. During grip tasks, only one of the two prime movers showed task-dependent changes. In the postural muscle AD there was no significant difference between MEPs during grip and grasp tasks; however, BB responses were larger during grasp than grip tasks. 4. MEPs simultaneously recorded in the prime movers were plotted against each other. The slope of the regression line for AD versus BB was larger in push than load tasks, whilst the changes in MEPs of 1DI and OP were independent during both grip and grasp tasks. 5. In three subjects, MEPs were also elicited by electrical stimulation during grip and grasp tasks. MEP changes tended to parallel those obtained for magnetic stimulation, but the increase in size of the electrically evoked MEPs during the precision task was smaller. 6. In all subjects the median and ulnar nerves were stimulated during grip and grasp tasks, and an H reflex was evoked in the hand muscles of five subjects. In no case did the two tasks produce reflexes of different amplitude. 7. The motor response of both proximal and distal muscles can be task dependent, in spite of the differences in their principal functional role and cortical representation. The modulation is related to the degree of control requested by the task, and is likely to reflect selective changes in the excitability of corticospinal neurones.