1. Electromyographic activity (EMG) is detectable in the feline stapedius muscle 6-10 ms after the onset of an intense sound presented to either ear. Stapedius reflexes evoked by ipsilateral and contralateral sound were measured electromyographically before and after brain stem lesions were made. In some cases, stapedius motor axons were cut; in others, brain stem regions containing motoneuron cell bodies were destroyed electrolytically. 2. Electrolytic lesions that contacted an anatomically separate cluster of stapedius motoneurons (the ventromedial perifacial group) greatly reduced responses to contralateral sound without noticeably affecting responses to ipsilateral sound. 3. Electrolytic lesions in other brain stem areas had different effects; one appeared to reduce responses to ipsilateral sound selectively, whereas others reduced both responses or had little effect. 4. After subsets of stapedius motor axons were cut at the facial colliculus in the floor of the fourth ventricle, responses to contralateral sound were almost eliminated, while substantial responses to ipsilateral sound remained. 5. The results are consistent with the hypothesis that inputs from the two cochleas are distributed inhomogeneously across the stapedius motoneuron pool in such a way as to produce a segregation of function, with motoneurons in one brain stem region responding preferentially (or exclusively) to contralateral sound and motoneurons in other regions responding preferentially (or exclusively) to ipsilateral sound. This topographic organization of acoustic input to the stapedius motoneuron pool produces a "central partitioning" in the acoustic stapedius reflexes similar in some respects to the partitioning observed in proprioceptive spinal reflexes.