Several studies have found a substantial association between subjects' judgments about the co-occurrence of personality items and the items' actual co-occurrence, lending support to the validity of implicit personality theory. This work has recently been called into question by an investigation of Mirels (1976) that reported a nonsignificant relationship between judged probabilities of item pairs and the pairs' empirical coendorsement probabilities (actually calculated conditional probabilities) derived from self-report responses. Using Mirels' data as well as new data, an appreciable and highly significant association between judgments of behavioral co-occurrence and empirical coendorsement was found, when the latter was measured by correlational-type indexes instead of conditional probability. This outcome indicates that the divergent results in the Mirels investigation were due to its use of the conditional probability index as a measure of empirical coendorsement. These findings add to the accumulating body of evidence pointing to the validity of implicit personality theory.