Objective: Independent evaluators (IE) are used widely in clinical trials to make unbiased determinations of treatment response. By virtue of being kept blind to treatment condition, however, IEs are also kept unaware of many pertinent clinical details that are relevant for decisions about clinical improvement. In this study, agreement among raters (children, parents, therapists, and IEs) about treatment response over the course of a 14-week clinical trial for pediatric obsessive-compulsive disorder (OCD) was examined in order to determine the utility of nonblind clinician and patient ratings of treatment response. Method: Participants were 71 youth (mean age = 12.2 years; 63.4% female) with a primary diagnosis of OCD and their parents participating in a psychotherapy trial. IEs provided response ratings (Clinician's Global Impressions-Improvement Scale; CGI-I) at Weeks 4, 8, and 14, and therapists, children, and parents completed independent CGI-I ratings at Weeks 2, 4, 8, and 14. Results: Nonlinear mixed models revealed differences in rating parties, with therapists and IEs slower to rate treatment improvement compared with children and parents, and growth curve models suggested that therapists and IEs produced generally consistent ratings. In addition, no evidence was found for an Evaluator x Treatment interaction, indicating that raters displayed these differences consistently across both active and placebo conditions. Conclusions: Youth and parents may be able to provide accurate ratings of global improvement; nonblinded treating clinicians (with training in research-oriented assessment) can offer global improvement ratings commensurate with blinded IEs. Findings suggest that alternatives (or additions) to the blinded-IE model may be appropriate for assessing global improvement, especially with the growing emphasis on dissemination and effectiveness trials. (Contains 1 table and 1 figure.)