Stimulation of the midbrain retrorubral (RRN), ventral paralemniscal tegmental field (vFTP), reticular tegmental (TRN), and pedunculopontine tegmental (PPN) nuclei was found to produce bilateral suppression of muscle tone in the unanesthetized, decerebrate animal. The RRN is the most rostral area shown to produce such suppression. This muscle tone suppression was frequency- and intensity-dependent. At low stimulus intensities, bilateral suppression was produced at these sites. At higher current and frequency levels, 2 types of muscle responses were found, excitation in PPN and RRN and initial suppression followed by excitation in TRN and vFTP. The mean latency to muscle tone suppression was not significantly different in TRN (36.8 msec) and RRN (36.5 msec). However, muscle tone suppression latency was significantly shorter in vFTP (31 msec) and PPN (27.1 msec). In addition to muscle tone suppression, stepping-like activity could be elicited at the same points by consecutive train stimulations in PPN and single train stimulation in TRN and vFTP. Thus, systems producing atonia are colocalized with those producing locomotion. We hypothesize that the midbrain atonia regions control more caudal regions producing muscle tone suppression in REM sleep, and that the locomotor and atonia eliciting regions are normally coactivated during REM sleep.