The electromyographic activity of flexors pollicis longus and brevis (with its synergists) has been compared on forcibly extending the thumb at various velocities with the muscles initially contracting. Both muscles gave short- and long-latency responses, but these differed in their relative magnitude with short-latency responses being better developed for the short flexor. With jerk-type stimuli both muscles gave short-latency responses with the expected slight difference in latency due to their different position in the arm. That of the long flexor was sometimes immediately followed by a long-latency response to the same stimulus. With slower displacements the short flexor regularly showed much more short-latency response than did the long flexor. The ensuing long-latency activity of the short flexor was normally appreciably less than that of the long flexor. However, since the short-latency response may be presumed to leave the motoneurones refractory it cannot be definitively concluded from this that acting in isolation long-latency pathways would be less potent for the short flexor, though this seems quite likely to be so. In some cases the first reflex activity occurred nearly synchronously for the two muscles in spite of their different separation from the spinal cord. That for the more distal short flexor was a short-latency response, whereas that for the more proximal long flexor was a long-latency response. The findings conflict with the provisional generalization that for muscles of the primate hand short-latency responses have been regularly supplanted by long-latency responses. They also provide the basis for a teleological argument against the view that the long-latency response is mediated transcortically.