After being exposed to either an aggressive or a nonaggressive filmed model, four- and five-year-old boys were tested for imitation while either alone or with a confederate who had also seen the film. The confederate's behavior was either inactive, the same as that in the film, the same as that in the film the child had not seen, novel but of the same category (aggressive or nonaggressive) as that in the film seen, or novel but in the category of the film not seen. These conditions conformed to the dimensions of a 2 × 6 factoral design. According to prediction, the imitation of subjects exposed to the aggressive model was greatest when the confederate's behavior was similar to the model's. No difference in imitation was produced by the confederate's actions if the boy had watched the nonaggressive model. The results were interpreted within the context of social comparison theory.