In this paper we describe two experiments in which the effects of typicality information from a target and a contrasting category on reaction times in a speeded categorization task were investigated. Experiment 1 elaborated on findings by Verbeemen et al. (2001) who found an effect of feature- and exemplar-based predictors from the target category, but virtually no effect of predictors from contrasting categories, in a wide range of categories. To avoid restrictions in the amount of relevant typicality information covered by these predictors we took a more direct approach. With directly rated typicality as a predictor for reaction times we obtained identical results. Experiment 2 expanded the set of concepts studied to superordinate animal concepts, again with typicalities as predictors. In these concepts we did find significant contributions of contrast categories. Possible explanations are discussed.