Three experiments identified conditions under which trait judgments made about a behavior were more likely to influence later judgments of the behavior. In Experiment 1, participants made trait judgments about numerous behaviors presented with photos of actors. Some behaviors were repeated, paired with the same or a different actor. All repeated behaviors were judged faster than new behaviors. Facilitation was greatest when repeated behaviors were paired with the same actor, suggesting greater influence of prior judgments in this condition. Experiments 2 and 3 replicated this effect, and the pattern of response times (RTs) suggested a stronger association between the actor and behavior when a prior impression of the actor had been formed (Experiment 2) and when the behavior was stereotypic of the actor's group (Experiment 3). Level of prejudice moderated RT patterns in Experiment 3. Implications for context effects, the nature of trait inferences, and stereotype change are discussed.