OBJECTIVE:The goal of this study was to investigate treatment outcome and mediators of cognitive-behavioral group therapy (CBGT) versus mindfulness-based stress reduction (MBSR) versus waitlist (WL) in patients with generalized social anxiety disorder (SAD). METHOD: One hundred eight unmedicated patients (55.6% female; mean age = 32.7 years, SD = 8.0; 43.5% Caucasian, 39% Asian, 9.3% Hispanic, 8.3% other) were randomized to CBGT versus MBSR versus WL and completed assessments at baseline, posttreatment/WL, and at 1-year follow-up, including the Liebowitz Social Anxiety Scale-Self-Report (primary outcome; Liebowitz, 1987) as well as measures of treatment-related processes. RESULTS: Linear mixed model analysis showed that CBGT and MBSR both produced greater improvements on most measures compared with WL. Both treatments yielded similar improvements in social anxiety symptoms, cognitive reappraisal frequency and self-efficacy, cognitive distortions, mindfulness skills, attention focusing, and rumination. There were greater decreases in subtle avoidance behaviors following CBGT than MBSR. Mediation analyses revealed that increases in reappraisal frequency, mindfulness skills, attention focusing, and attention shifting, and decreases in subtle avoidance behaviors and cognitive distortions, mediated the impact of both CBGT and MBSR on social anxiety symptoms. However, increases in reappraisal self-efficacy and decreases in avoidance behaviors mediated the impact of CBGT (vs. MBSR) on social anxiety symptoms. CONCLUSIONS: CBGT and MBSR both appear to be efficacious for SAD. However, their effects may be a result of both shared and unique changes in underlying psychological processes.