Fifteen years ago, W. Davison introduced the third-person effect hypothesis, that individuals believe they are less influenced than others by media messages. Although third-person effect is a perceptual bias, Davison believed that individuals act on such misperceptions. Few studies since have tested the behavioral aspect of the third-person effect. In addition, previous studies reporting differences in third-person effect due to message type (i.e. Public Service Announcements (PSAs) vs. advertisements) were controlled to isolate the effects of message type from content and context. The current study sought (1) to document third-person effect among minority "at-risk" youth within the context of safer sex messages; (2) to determine the differences in third-person effects (if any) between PSAs and advertisements with similar content; and (3) to determine the link (if any) between third-person effect and risky sexual behaviors among youth. Subjects were 177 adolescents ranging in age from 8 to 17 enrolled in programs servicing "at-risk" youth in urban New Jersey. Findings indicate that third-person effect is an appropriate framework for understanding how "at-risk" youth perceive safer sex campaigns. The study also confirms the behavioral aspect of third-person effect by linking it with sexual risk behaviors. No difference in third-person effect was found as a result of different message types. Contains 25 references. (Author/RS)