1. During maintained ankle extension with background torques ranging from 0 to 70 N.m, the ankle extensors were stretched by a 5 degree rotation of the ankle joint. Maximal stretch velocity was 170 degrees/s. Regression analysis of simultaneous measurements of total torque and needle and surface electromyograms (EMG) from the soleus and gastrocnemius muscles showed that the soleus muscle generates about two-thirds of the maximal torque (approximately 120 Nm) with the subjects in sitting position. In addition, it was found that there is considerable cross talk between the soleus and gastrocnemius muscles when EMGs are recorded by surface electrodes. 2. The soleus EMG response to stretch began with a "phasic" response (latency 41 +/- 4 ms, mean +/- SD), consisting of two peaks, labeled M1 and M2. The phasic response ended 120-140 ms after stretch onset and was followed by a period of reduced EMG activity, ending at 170-210 ms. After this "silent period," a smaller "tonic" response was seen. The phasic responses of the soleus muscle were much larger than the corresponding responses in the anterior tibial muscle. In contrast, the tonic responses were comparable in the soleus and anterior tibial muscles. 3. The amplitudes of the phasic M1 and M2 responses were independent of the level of the background contraction. This disagrees with the "automatic gain principle," according to which the amplitudes of M1 and M2 should increase proportionally with the background EMG. In contrast to the phasic responses, the amplitude of the tonic EMG response, measured 200-400 ms after stretch onset, followed the automatic gain principle.(ABSTRACT TRUNCATED AT 250 WORDS)