In the present study, subjects had to generate an evaluative judgment about a target person on the basis of his behaviour that had both positive and negative implications. In a previous phase of the study that was ostensibly unrelated to the judgment task, the relevant trait categories were primed. Subsequently, half of the subjects were reminded of the priming episode. Consistent with earlier research (e.g. Lombardi, Higgins and Bargh, 1987; Newman and Uleman, 1990) that used memory of the priming events as a correlational measure, a contrast effect was found under the ‘reminding’ condition and assimilation resulted when subjects were not reminded of the priming episode. This pattern of results is interpreted as the consequence of corrective influences.