1. On the basis of the orderly activation of motoneurons in a pool, one would predict that motor unit activity and whole muscle force will change at least roughly in parallel: active motor units should continue to fire as net muscle force increases and quiescent motor units should remain inactive as muscle force decreases. We have consistently observed this relationship in our studies of the medial gastrocnemius (MG) muscle, but here we report an uncoupling of the soleus muscle and some of its motor units. 2. Physiological properties and firing behaviors of 20 soleus motor units were characterized in five decerebrate cats with the use of intra-axonal stimulation and recording. Motor unit firing was elicited in reflexes initiated by muscle stretch, nerve stimulation, and mechanical stimulation of the heel. Particular emphasis was placed on the heterogenic reflexes produced in soleus by ramphold-release stretches of the MG muscle. In agreement with previous reports, either net heterogenic excitation or inhibition of the soleus muscle was produced in separate trials of MG stretch. 3. During excitation of soleus in autogenic stretch reflexes and in crossed-extension reflexes, all 20 units were recruited or increased firing, i.e., unit firing was coupled with soleus force. In the other reflexes, however, unit firing and muscle force were uncoupled for 10 of these units. Six tonically active motor units were inhibited during an increase in soleus force produced by MG stretch or by mechanical stimulation of the heel. Four motor units were activated during a decrease in soleus force produced by the same stimuli. 4. Six motor units were studied during both soleus inhibition and excitation evoked by MG stretches. One motor unit was consistently coupled to the soleus muscle response; firing increased during soleus excitation and decreased during inhibition. However, four soleus motor units were inhibited under both conditions, and one unit was excited under both conditions. Thus the firing behavior of five of these six motor units was the same in response to MG stretch, irrespective of the soleus response. 5. The uncoupling was most clearly recognized when tonically active units ceased firing during net excitation of the soleus muscle and when silent units began firing during net inhibition of the soleus muscle. Unit responses were not as striking in all trials of MG stretch (spike number increased or decreased relative to prestretch values by 1-4 spikes), but the responses were consistent across trials; in multiple stretches, spike number commonly either increased or decreased. Intertrial regularity was also observed in units for which firing was coupled with the net reflex response of the soleus muscle. 6. Divergence in the firing of soleus motor units was also observed in three cases in which records were taken simultaneously from two motor units. In one pair, one unit increased and the other decreased firing during MG stretch-evoked inhibition of soleus. In the other two pairs, one unit increased and the other decreased firing when soleus was excited by heel stimulation. In all pairs, the unit that decreased firing under these conditions had the lowest recruitment threshold in response to the soleus stretch. 7. Although all soleus motor units were classified as slow-twitch (type S), variation in their physiological properties bore some relation to firing behavior. Those units recruited during periods of soleus inhibition exhibited among the fastest conduction velocities and contraction times in our sample. In all three unit pairs sampled, the unit expressing decreases in firing had the slower conduction velocity and contraction time. 8. These findings demonstrate that soleus motor units are differentially activated and deactivated by peripheral afferents. (ABSTRACT TRUNCATED)