Several strategies are proposed as bases for judgments of covariation between events. Covariation problems were structured in such a way that patterns of correct and incorrect judgments would index the judgment rule being used by a given subject. In two experiments, 10th-grade or college subjects viewed a set of covariation problems, each of which consisted of a set of observations in which each of two events was defined as present or absent. Subjects were asked to identify the relationship between the events. Subjects' response patterns suggested that the modal strategy was to compare frequency of confirming and disconfirming events in defining the relationship. Response accuracy was influenced by pretraining on the concept of covariation and by response format. Instructions to sort the observations did not influence judgment accuracy. © 1980 Psychonomic Society, Inc.