The electromyographic activity of flexor pollicis longus has been recorded in normal human subjects on moving the tip of the thumb with the proximal phalanx clamped. Ramp and hold displacements (stretches) were compared with high-frequency sinusoidal movement (vibration). The subject exerted a constant flexor force between stimuli and made no voluntary response to them. On stretching the muscle by forcibly extending the thumb at various constant velocities the usual combination of short-latency (ca. 25-30 ms) and long-latency (ca. 40 ms) components of response were observed. The short-latency response progressively predominated as the velocity was increased (60-900 deg s-1, 9 deg joint displacement). One subject still showed only a long-latency response with the fastest stretch, arguing that it is a distinct reflex entity. On commencing vibration (143 Hz, 3 deg movement peak-to-peak) a short-latency response was regularly obtained, but any long-latency response was always small in relation to that elicited by stretch. This was equally so when the short-latency responses to the two types of stimulation were matched by using appropriate parameters of stimulation. The time course of the vibration response did not change appreciably with change of amplitude of vibration, so that its temporal profile was always quite different from that of the stretch response. The observed differences are in accordance with the hypothesis that the spindle group II afferents produce the long-latency excitation, with the time lost peripherally in afferent conduction rather than centrally. In relation to the strength of their Ia excitatory actions, stretch is known to excite secondary afferents more powerfully than does vibration. The findings are not readily accommodated on the hypothesis that the long-latency response is a transcortical reflex elicited by the initial Ia input, since vibration should then also have had a powerful long-latency action. Similar responses to vibration were obtained when it was applied percutaneously to the tendon of flexor pollicis longus 6 cm above the wrist. Also, those elicited by thumb vibration persisted largely unchanged when the thumb was anaesthetized. This confirms that they were dependent upon the excitation of receptors in flexor pollicis longus, presumably the Ia afferents, rather than upon cutaneous or joint receptors in the thumb. The stretch responses also depended upon muscle receptors, since they too survived anaesthesia.(ABSTRACT TRUNCATED AT 400 WORDS)