Social anxiety disorder (SAD) is associated with considerable individual suffering and societal costs. Although there is ample evidence for the efficacy of cognitive behavior therapy, recent studies suggest psychodynamic therapy may also be effective in treating SAD. Furthermore, Internet-based psychodynamic therapy (IPDT) has shown promising results for addressing mixed depression and anxiety disorders. However, no study has yet investigated the effects of IPDT specifically for SAD. This paper describes a randomized controlled trial testing the efficacy of a 10-week, affect-focused IPDT protocol for SAD, compared with a wait-list control group. Long-term effects were also estimated by collecting follow-up data, 6, 12, and 24 months after the end of therapy. A total of 72 individuals meeting diagnostic criteria for DSM–IV social anxiety disorder were included. The primary outcome was the self-report version of Liebowitz Social Anxiety Scale. Mixed model analyses using the full intention-to-treat sample revealed a significant interaction effect of group and time, suggesting a larger effect in the treatment group than in the wait-list control. A between-group effect size Cohen’s d = 1.05 (95% [CI]: [0.62, 1.53]) was observed at termination. Treatment gains were maintained at the 2-year follow-up, as symptom levels in the treated group continued to decrease significantly. The findings suggest that Internet-based affect-focused psychodynamic therapy is a promising treatment for social anxiety disorder.