Accuracy in psychiatric diagnosis is critical for evaluating the suitability of the subjects for entry into research protocols and for establishing comparability of findings across study sites. However, training programs in the use of diagnostic instruments for research projects are not well systematized. Furthermore, little information has been published on the maintenance of interrater reliability of diagnostic assessments. At the UCLA Research Center for Major Mental Illnesses, a Training and Quality Assurance Program for SCID interviewers was used to evaluate interrater reliability and diagnostic accuracy. Although clinically experienced interviewers achieved better interrater reliability and overall diagnostic accuracy than neophyte interviewers, both groups were able to achieve and maintain high levels of interrater reliability, diagnostic accuracy, and interviewer skill. At the first quality assurance check after training, there were no significant differences between experienced and neophyte interviewers in interrater reliability or diagnostic accuracy. Standardization of training and quality assurance procedures within and across research projects may make research findings from study sites more comparable.