Objectives. There is much debate as to whether the treatment effects achieved in well-controlled studies such as randomized controlled trials (RCTs) are generalizable to more 'naturalistic' clinical populations, such as that seen in private practice. The current study sought to examine this issue in relation to social phobia. Design. A benchmarking strategy was used to compare the effectiveness of a cognitive-behaviour therapy group programme for social phobia that was developed and evaluated in a research unit, to that of a private practice population. Methods. Fifty-eight participants from a university research unit and 54 participants from an independent private practice who met the principal diagnostic criteria for social phobia completed the 10-session group programme. Symptom severity was measured at pre-treatment, post-treatment, and 3 months after treatment. Results. No significant treatment differences were found between the research unit and private practice groups. Both groups showed significant treatment effects that were maintained at 3-month follow-up. Conclusion. These findings suggest that treatments developed for RCTs are potentially transportable to private practice settings.