A common assumption in perceptual learning is that the improvement of performance basically depends on the amount of training. However, other factors such as sleep and training intensity (trials per session) have also been shown to be important. We trained four groups with 1600 trials in total, using a Chevron discrimination task. Groups differed in the number of trials per session and the amount of sleep between sessions. One group trained two sessions on two consecutive days with 800 trials per session. Performance improved. Another group trained ten sessions on ten consecutive days with 160 trials per session. Performance did not improve. This result suggests that a minimal number of trials per session is necessary for perceptual learning. Next, we calculated the changes in performance between sessions for the groups which improved performance. No significant improvement of performance was found irrespective of the amount of sleep between sessions. Hence, perceptual learning with Chevrons seems to require a minimal number of trials per session, but no sleep