To examine the influence of conditioned fear stimuli on sleep-wake states, we recorded sleep in Sprague–Dawley rats after exposure to tones previously paired with footshock. After habituation to a recording chamber and the recording procedure, a baseline sleep recording was obtained the next day. One day later, experimental animals were exposed to shock training designed to induce conditioned fear (FC), consisting of five tone-footshock pairings. The 5-s tones (conditioned stimuli; CS) co-terminated with 1-s footshocks (unconditioned stimuli; US). The next day sleep was recorded for 4 h in the recording chamber after presentation of five CSs alone. Sleep efficiency (total sleep time/recording period) and REM sleep (REM) and non-REM (NREM) measures were determined. While sleep efficiency was not significantly changed after CS presentation, the percentage of total sleep time spent in REM (REM percentage) was reduced in the FC animals. The reduction in REM percentage in the FC animals was due to a decrease in the number of REM bouts. In a separate experiment, we repeated the procedures, except the tones and shocks were presented in an explicitly unpaired (UP) fashion. The next day, presentation of the tones increased REM percentage in the UP group. Results are discussed in terms of the decreases in REM as a response to conditioned fear, and the relevance of these findings to the sleep changes seen in post-traumatic stress disorder (PTSD).