The results of 4 experiments suggest that the role of categorization in comparative judgment is much greater than previously believed. Ss judged which of 2 objects, selected from 2 taxonomic categories, was larger. Items from one category were always larger than items from the other category. When the items in the pair were from different categories, the semantic distance effect was attenuated if the relation between the categories was also generally true in the real world. When the experimental relation between the 2 categories differed from that of the real world, no diminution of the magnitude of the distance effect was observed. Further experiments rule out an artifactual explanation and also establish that diminution of the distance effect will occur even if the correlation between category membership and magnitude is relatively modest.