This article examines the development of inductive generalization, and presents a model of young children's induction and two experiments testing the model. The model specifies contribution of linguistic labels and perceptual similarity to young children's induction and predicts a correspondence between similarity judgment and induction of young children. In Experiment 1, 4- to 5-year-olds, 7- to 8-year-olds, and 11- to 12-year-olds were presented with triads of schematic faces (a Target and two Test stimuli), which varied in perceptual similarity, with one of the Test stimuli sharing a linguistic label with the Target, and another having a different label. Participants were taught an unobservable biological property about the Target and asked to generalize the property to one of the Test stimuli. Although 4- to 5-year-olds’ proportions of label-based inductive generalizations varied with the degree of perceptual similarity among the compared stimuli, 11- to 12-year-olds relied exclusively on labels, and 7- to 8-year-olds appeared to be a transitional group. In Experiment 2 these findings were replicated using naturalistic stimuli (i.e., photographs of animals), with perceptual similarity manipulated by “morphing” naturalistic pictures into each other in a fixed number of steps. Overall results support predictions of the model and point to a developmental shift from treating linguistic labels as an attribute contributing to similarity to treating them as markers of a common category—a shift that appears to occur between 8 and 11 years of age.