The associationistic view of person perception states that people are perceived in terms of trait covariations. The dimensional view maintains that others are perceived by means of a limited number of dimensions. In contrast, the typological view is that others are perceived in terms of person types, and traits within a given person type have a unique interactive relation. In Experiment l, associationistic, dimensional, and typological representations of implicit personality theories were empirically derived. The derived stimuli were used in Experiment 2, which examined the effects of person type membership on impression priming and perceived trait belongingness, controlling for associationistic and dimensional factors. As expected, results validated the unique contribution of a typological approach to person perception. Implications of the present findings for the implicit personality theory, stereotype, and person memory literatures are discussed. People constitute important categories of objects. It is of no surprise, then, that thinking about people is such a pervasive phenomenon. We routinely form impressions of other people's personalities, make predictions about their behavior, wonder about their intentions, and evaluate their performance outcomes. In this article, we are concerned with the issue of how we think about, perceive, or cognitively represent other people. How Do People Think About People? There are at least three general views that speak to this issue: The associationistic, the dimensional, and the typological view. i Associationistic and Dimensional Views The simplest view, the associationistic, is that people think about others in terms of trait covariations. Expectations about a target person's standing on some unknown trait (e.g., shyness) are derived from beliefs about the correlations of that trait with known traits (e.g., outgoingness, aloofness). More formally, Trait A is expected to accurately describe a person who has Traits B, C, and D to the extent that A is believed to correlate positively with B, C, and D. The expectation for a given unknown trait depends on the perceived relations between that trait and all We thank the participants in the Nags Head Conference on Social Cognition, especially Dave Schneider, for their helpful suggestions. We also thank Rob Nosofsky and the anonymous reviewers for their perceptive comments on drafts of this article and Phil Wood for his expert advice on multidimensional scaling and cluster analysis. The help of Mary Finch, Deborah Weaver, and Lee Metzler in conducting this research is gratefully acknowledged.