Background: This study evaluates the benefits of a self-directed Internet intervention for depression (MoodGYM) delivered as a part of the high school curriculum. Method: One hundred and fifty-seven girls, aged 15 and 16 years, were allocated to undertake either MoodGYM or their usual curriculum. MoodGYM's impact on depressive symptoms, risk of depression, attributional style, depression literacy and attitudes toward depression was examined using random effect regression. Results: MoodGYM produced a significantly faster rate of decline in depressive symptoms over the trial period than the control condition. The effect size for MoodGYM was not significant immediately after the intervention (Cohen's d 5 .19, 95% CI À.18-.56) but was moderate and significant 20 weeks after the intervention (d 5 .46, 95% CI .10-.82). Girls with high depression scores before intervention showed the strongest benefits on self-reported depression at follow-up (d 5 .92, 95% CI .10-1.38). There were no significant intervention effects on depression status, attributional style, depression literacy, and attitudes. Approximately 70% of girls in the MoodGYM group completed less than three of its modules and completion of fewer modules was related to high depression score before intervention. Conclusions: The findings suggest that there are benefits from MoodGYM on self-reported depressive symptoms but has low rates of completion highlight problems in ensuring adherence to Internet programs for depression. Depression and anxiety 26:65-72, 2009.