Following one adaptation night, the nocturnal sleep of nine female college students was recorded for two baseline and two experimental nights. In both experimental nights, prior to sleep the subjects performed a 4 hour long visual task requiring the recognition of letters tachistoscopically presented in the right visual hemifield. In the first experimental night eye movements were permitted, while in the second subjects were allowed only covert movements of attention while keeping their gaze on a fixation point. Results show that, as a consequence of the experimental manipulation, there is an increase of REM duration and a shortening of REM latency without any concomitant significant change in REM density and in the duration and latency of the other sleep stages. The increase of REM duration was greater in the night in which only attentional "movements" were permitted (i.e., when there is a covert orienting of attention), suggesting that a load of spatial attention might have a role in increasing REM pressure.