Cognitive Behavioural Group Therapy (CBGT) for social phobia has been shown to be efficacious within research units and effective within a variety of real world clinical settings. However, most effectiveness studies of CBGT for social phobia have (a) used protocols without demonstrated efficacy, (b) not included direct comparison groups, and/or (c) contained features of efficacy trials. This study addressed these limitations by using a benchmarking strategy to compare outcomes from the same CBGT protocol used in both a research unit and a community clinic. Research (N = 71) and community (N = 94) patients completed the same 12-session protocol, which resulted in significant reductions in social anxiety and life interference at post-treatment. Compared to research unit patients, community patients had more severe symptoms and life interference at pre-treatment, and were more likely to be male, use medication, have comorbid disorders, and have lower educational attainment. Importantly, degree of improvement on social anxiety symptoms and life interference did not differ across the treatment settings for either completer or intention-to-treat analyses. There was some evidence that being younger, single, and having a depression diagnosis were associated with dropout. Pre-treatment symptoms and number of diagnoses predicted post-treatment symptoms. Consistent with previous uncontrolled trials, it is concluded that CBGT is effective within community mental health clinics.