This article critically reviews the research literature on peer modeling among children as a function of model attributes. Peer modeling is hypothesized to depend in part on perceived similarity between model and observer. Similarity serves as an important source of information for gauging behavioral appropriateness, formulating outcome expectations, and assessing one's self-efficacy for learning or performing tasks. Research is reviewed on the effects of model age, model sex, model competence, number of models, and model background. Peer models can foster diverse types of behavioral change in children, but attribute similarity does not automatically enhance modeling. The conditions under which similarity promotes behavioral change are discussed. Future research needs to assess children's self-perceptions, as well as maintenance and generalization of behavioral changes. It is suggested that classroom peers can help train social skills, enhance self-efficacy, and remedy skill deficiencies. Article: Modeling is an important means of acquiring skills, beliefs, and novel behaviors (Bandura, 1986; Rosenthal & Zimmerman, 1978). In school, teachers serve as powerful models for children, but children also are exposed to many peers who are the same age and who often are similar in other ways (e.g., sex, competence). Peer models can affect many aspects of children's lives (Hartup & Lougee, 1975). The purpose of this article is to critically review the literature on peer modeling among children as a function of various model attributes. As used throughout this article, peer denotes a child who is roughly equivalent in development to the observer, model is an individual whose behaviors, verbalizations, and expressions are attended to by the observer and serve as cues for subsequent modeling, and modeling refers to behavioral change that derives from observing others (Berger, 1977; Field, 198 1; Hartup, 1978).