To determine whether the white rat can learn to make a one-trial discrimination reversal, nine albino rats, 100-140 days old were given a series of ten position discrimination problems in which they learned to go to one of two goal boxes for a food reward. Each animal was first run against his position preference. When this problem was mastered, the problem was reversed (e.g., to the left first, then right, then the left box again became positive, and so on throughout the series of ten problems). The problem was reversed after the animal had made 11 out of 12 responses correct, with the last 8 consecutively correct. All of the rats learned to make a one-trial discrimination reversal by going to the previously negative food box after a single nonreinforced response to the previously positive one. It is suggested that the critical variable making such learning possible is the use of a criterion of learning.