Extensive evidence has linked both paradoxical sleep (PS) and stress to memory processing. The purpose of the present study was to examine the effect of social instability stress on memory and to verify whether this stress interferes with the amnesic effect of PS deprivation using the modified multiple platform method. In addition to the PS-deprived group (put onto narrow platforms inside the deprivation tanks) two control groups were used: one of them remained in its home-cages and the other was placed inside the deprivation tanks, onto a grid that contained large platforms on it. All groups were subdivided in socially stable and unstable conditions. Immediately after 96 h of sleep deprivation, the animals were trained in three different memory tasks: inhibitory avoidance, classical fear conditioning to a discrete stimulus and contextual fear conditioning. Twenty-four hours after training, the animals were tested in order to assess task acquisition. The results showed that social instability did not impair the performance of animals nor interacted with PS deprivation in any of the tasks. Grid control animals presented a selective impairment in the inhibitory avoidance task and contextual, but not in the classical, fear conditioning task, compared to cage control rats. This finding could be due to the stress to which grid control animals were exposed (humidity and luminosity) during the manipulation period. PS-deprived animals exhibited poorer performance than the other groups in all tasks. As they also showed an increased threshold to shock-induced vocalisation, but not to flinch response, it is not possible to completely rule out a decreased response to noxious stimulation as a contributing factor for the present results with PS deprivation.