This experiment was designed to examine the development of a spatial learning set in rats and some of the variables influencing the retention of individual problems. The apparatus was a plus maze. At the beginning of each test, the rat was put on two arms, each in a different place. Food was present in one of the arms, but not in the other. The rat was then given a choice between these two places; the correct response was to return to the place that previously contained food (win-stay, lose-shift, response-reinforcement contingency). Fifty different two-choice spatial discriminations were given, each in a different location. At the end of testing, the mean percentage of correct responding for the first choice between the two places was 83%. Control procedures showed that the discriminative stimuli were distal, extramaze spatial stimuli. Variations of the procedure examined the influence of proactive interference and temporal delay on the memory for each discrimination. These results demonstrate that rats can develop a spatial learning set and provide new information about the characteristics of the memory underlying learning sets.