The role of posttraining paradoxical sleep (PS) in spatial or nonspatial learning in the Morris water maze was evaluated. Sprague-Dawley rats were given a 12-trial training session in either the hidden or the visible platform versions of the task. Subgroups then underwent paradoxical sleep deprivation (PSD) beginning at different times after training. Rats with PSD imposed from 1-4 hr after spatial training had poorer retest scores than any other group. Other rats, implanted with electrodes to permit continuous recording of sleep electroencephalography, were found to undergo a prolonged period of elevated PS after spatial training. By contrast, rats trained in the nonspatial version of the water maze task did not show retention deficits after PSD or elevated PS after training. These results support a role for PS in spatial, but not nonspatial, learning in the Morris water maze.