Sleep data were obtained on 11 patients who had survived traumatic events and who complained of sleep disturbances. Each was awakened from REM and non-REM sleep for dream recall. The patients had lower sleep efficiency indices (because of prolonged sleep latency and larger amounts of "awake" plus "movement" time within sleep periods), shorter REM time, and longer REM latencies than did control subjects. Four of the 11 patients had REM- and non-REM-related nightmares, which, in two sea disaster patients, were associated with REM-related motor activity. The rest of the patients had unusually low dream recall in spite of high eye movement density.