Rats were subjected to a partial, selective paradoxical sleep (PS) deprivation. Two groups were trained in a shuttle avoidance task, one receiving posttraining amygdaloid stimulations (AS). A third group received AS, but was not trained. Levels of PS in the AS trained group were higher than in the AS non-trained controls, while the reverse was true for the slow wave sleep (SWS) measure. Both AS groups had levels of PS and SWS superior to the non-AS trained rats. In a second experiment, a group of normally rested AS animals was trained in the same task with a non-AS control group. These groups did not differ in terms of learning performance. Results were believed to support the hypothesis of a close relationship between sleep and learning.