A survey of the literature on the relation between spontaneous activity and maze learning pointed to the absence of any relationship between the two traits. In order to test the supposition that these negative results were due to errors of measurement, the present experiment was undertaken. The animals used came from the F17 generation of strains selected on the basis of activity. Since the two strains represent relatively homogeneous groups at the extremes of activity, errors of measurement would presumably be of less significance. The active animals proved to be significantly better learners than the inactive, using errors as the criterion of learning. No difference in time is evident, after allowing for the different number of errors made by the strains. The probability that errors of measurement have been responsible for previous negative results is indicated not only by the fact that they are of less significance in this experiment but also by the greater differences in learning observed under conditions of greater incentive and the influence of a greater reliability of the maze on the significance of the difference in learning between the two strains. The evidence suggests that the two strains differ in motivation, and that this difference is in part responsible for the difference in maze learning. There is a differential effect of changing the amount of the incentive, the active strain being quite sensitive to changes, the inactive being quite insensitive. Certain implications of the findings are discussed and the possibility that the common factor underlying maze learning and spontaneous activity is basal metabolism is pointed out. (PsycINFO Database Record (c) 2006 APA, all rights reserved)