Research on causal-based categorization focuses on how people categorize exemplars that have causally linked features. One prominent account, the generative model (Rehder, 2003) models membership judgments as a function of an exemplar's likelihood being generated by the category's causal model. In contrast, Mayrhofer and Rothe (2012) found that the explanatory role of the causal model strongly influences membership judgments - indicating the importance of explanatory reasoning processes and that in such tasks people might be guided by explanatory goodness (i.e., how well an exemplar's membership can be explained in the light of the category's causal relations). However, the evidence for this claim was quite indirect so far. In the present categorization study, we collected judgments about category membership, frequency, and explanatory goodness. In contrast to the predictions of the generative model, membership ratings are far better resembled by ratings of explanatory goodness than by subjects' estimations of exemplar likelihood.