Scientific evaluation of the efficacy of treatments for mental disorders is an undertaking of critical importance. However, the design of trials evaluating psychosocial treatments may determine whether or not they will ultimately provide meaningful data that may guide further treatment development. In this paper, we address areas of methodological concern in the evaluation of psychosocial treatments. These are discussed in the context of psychosocial treatments for social phobia, whether they be the sole focus of an outcome trial, compared to other psychosocial or pharmacological treatments, or combined with pharmacotherapies. We address issues of patient representativeness, adequacy of treatment as conducted, selection and design of control conditions, and assessment of change over the course of treatment for social phobia. Most of these issues have implications for the evaluation of pharmacotherapies for social phobia as well.