Guided by the Behavioral Vaccine Theory of prevention, this study uses a no-control group design to examine intervention variables that predict favorable changes in depressive symptoms at 6- to 8-week follow-up in at-risk adolescents who participated in a primary care, Internet-based prevention program. Participants included 83 adolescents from primary care settings ages 14 to 21 (M = 17.5, SD = 2.04), 56.2% female, with 41% non-White. Participants completed self-report measures, met with a physician, and then completed a 14-module Internet intervention targeting the prevention of depression. Linear regression models indicated that several intervention factors (duration on website in days, the strength of the relationship with the physician, perceptions of ease of use, and the perceived relevance of the material presented) were significantly associated with greater reductions in depressive symptoms from baseline to follow-up. Automatic negative thoughts significantly mediated the relation between change in depressive symptoms scores and both duration of use and physician relationship. Several intervention variables predicted favorable changes in depressive symptom scores among adolescents who participated in an Internet-based prevention program, and the strength of two of these variables was mediated by automatic negative thoughts. These findings support the importance of cognitive factors in preventing adolescent depression and suggest that modifiable aspects of technology-based intervention experience and relationships should be considered in optimizing intervention design.