Small, rapid stretches were applied to the soleus muscle during the stance phase of walking by lifting the forefoot with a pneumatic device. Stretch responses were induced in the soleus muscle by the disturbance. The amplitude and time course of the responses from the soleus muscle were a function of both the kinematics of the disturbance and the time in the step cycle when the disturbance was applied. The step cycle was divided into 16 equal time parts, and data obtained within each of these parts were averaged together. The electromyographic (EMG) response of the soleus muscle showed a time course that was similar to the time course of the angular velocity induced by the disturbance at the ankle. Three linear equations were used to predict the EMG response from the soleus muscle as a function of the angular kinematics of the disturbance: 1) velocity, 2) velocity and displacement, 3) velocity, displacement and acceleration. Introduction of a pure delay between the EMG and the kinematics substantially improved the predictions. Most of the variance (70%) in the EMG response could be accounted for by the velocity of the disturbance alone with an optimal delay (average 38 ms). Inclusion of a displacement term significantly increased the variance accounted for (85%), but further addition of an acceleration term did not. Since the velocity of the disturbance accounted for most of the variance, the reflex gain was estimated from the velocity coefficient. This coefficient increased in a ramp-like fashion through the early part of the stance phase, qualitatively similar to the increase in the H-reflex. Based on these identified gains, this reflex pathway was estimated to contribute substantially (30% to 60%) to the activation of the soleus muscle particularly during the early part of the stance phase.