To test the hypothesis that a transcortical reflex contributes to the stretch-evoked long-latency electromyographic (e.m.g.) response we documented the responses of identified corticomotoneuronal (c.m.) cells and their target muscles to perturbations of active wrist movements. Macaque monkeys performed ramp-and-hold wrist movements against elastic loads, alternating between flexion and extension zones; brief (25 ms) torque pulses were intermittently applied during the hold period. C.m. cells were identified by a clear post-spike facilitation in spike-triggered averages of forelimb muscle e.m.g. activity. Activity of c.m. cells and twelve wrist and digit flexor and extensor muscles was recorded during: (a) active ramp-and-hold wrist movements, (b) passive ramp-and-hold wrist movements, and (c) torque perturbations applied during the hold phase of active flexion and extension which either lengthened or shortened the c.m. cell's target muscles. Muscle-lengthening perturbations evoked a reproducible pattern of average e.m.g. activity in the stretched muscles, consisting of two peaks: the first response (M1) had an onset latency of 11.2 +/- 2.1 ms (mean +/- S.D.), and the second (M2) began at 27.9 +/- 5.1 ms. Torque perturbations which shortened the active muscles also evoked a characteristic e.m.g. response consisting of an initial cessation of activity at 13.5 +/- 3.4 ms followed by a peak beginning at 33.9 +/- 3.0 ms. The responses of twenty-one c.m. cells which facilitated wrist muscles were documented with torque pulse perturbations applied during active muscle contraction. Twenty of twenty-one c.m. cells responded at short latency (23.4 +/- 8.8 ms) to torque perturbations which stretched their target muscles. For each c.m. cell-target muscle pair, transcortical loop time was calculated as the sum of the onset latency of the c.m. cell's response to lengthening perturbations (afferent time) and the onset latency of post-spike facilitation (efferent time). The mean transcortical loop time was 30.4 +/- 10.2 ms, comparable to the mean onset latency of the M2 peak (27.9 +/- 5.1). The duration of a c.m. cell's response to torque perturbations provides a further measure of the extent of its potential contribution to the M2 muscle response. In all cases but two, the c.m. cell response, delayed by the latency of the post-spike facilitation, overlapped the M2 e.m.g. peak.