Common factors are understood to play an important role in many therapeutic interventions. What is not yet clear is whether the relative importance of common factors in therapeutic interventions varies as a function of treatment type and/or disorder. In this study, we were specifically interested in the common factor of working alliance. We randomized adults with generalized social anxiety disorder (SAD) to 12 weeks of small group cognitive behavioral therapy (CBT; N = 46, 58.7% female, average age 33.41 ± 7.62 years) and 12 weeks of small group mindfulness-based stress reduction (MBSR; N = 40, 57.5% female, average age: 32.77 ± 7.94). We measured social anxiety clinical symptom severity using the Liebowitz Social Anxiety Scale-Self-Report (LSAS-SR) scale at baseline, post-treatment, 6 months, and 12 months post-treatment. We also measured the participant’s self-reported working alliance with their therapist/instructor at session 5. Results indicated that in the CBT group, working alliance measured at session 5 was not predictive of clinical symptom severity (LSAS-SR) at post-intervention (p > .27), 6 months (p > .39), or 12 months (p > .24) post-treatment. By contrast, in the MBSR group, working alliance measured at session 5 was predictive of clinical symptom severity at post-intervention (∆R2 = .11), 6 months (∆R2 = .15), and 12 months (∆R2 = .18) post-treatment. Follow-up analyses indicated that when compared to CBT, greater Working Alliance Inventory during MBSR was meaningful in that it predicted long-term (12-month follow-up) social anxiety symptom reduction (∆R2 = .12). These findings suggest that the importance of working alliance may vary by type of intervention, at least in the case of SAD.