The article focuses on three issues: (1) Do parental and adolescent reports of delinquent peers reflect the same underlying construct? (2) If not, how does the effect on delinquency of a measure of delinquent peers based on parental reports compare with that of a measure based on adolescents’ reports when the two measures are included in the same equation? and (3) Given possible contamination in measurement, can respondents’ reports of peer delinquency be regarded simply as additional indicators of self-reported delinquency? Structural equation modeling was used to address these issues with data from the National Youth Survey. Parental and adolescent reports of delinquent peers did not reflect a single latent construct. Further, the measure of delinquent peers based on parental reports yielded a significant effect on subsequent self-reported delinquency (controlling for self-reports of prior delinquency) whereas the measure based on adolescent reports had no significant effect. There was appreciable (but not complete) overlap between adolescent-reported peer delinquency and self-reported delinquency. Tables, figures, notes, references