It is unclear whether deviant sexual preferences distinguish adolescents who commit sex offenses in the same way that such deviance characterizes adult sex offenders. We compared male adolescents (mean age = 15 at the time of a referral sex offense), matched adult sex offenders, and normal men (adult nonoffenders or nonsex offenders). We hypothesized the following: phallometric responses of the adolescents would be similar to those of adult sex offenders and would differ from normals; adolescents with male child victims would exhibit greater evidence of sexual deviance than those whose only victims were female children; among adolescents who had molested children, those with a history of sexual abuse would exhibit more evidence of sexual deviance than those with no such history; and phallometric measures would predict recidivism. With some notable exceptions or qualifications, results confirmed the hypotheses. Phallometry has valid clinical and research uses with adolescent males who commit serious sex offenses.