The Structured Clinical Interview for DSM-III-R was used to examine the effects of the co-occurrence of psychiatric and substance dependence disorders on diagnostic reliability. The test-retest reliability over a 1-week period was studied in groups of: a) individuals with current substance abuse diagnoses (N = 97), b) individuals with past, but not current, drug histories (N = 146), and c) individuals without substance abuse diagnoses (N = 356; primarily psychiatric patients). A measurement of reliability (Kappa coefficients) was estimated for four general psychiatric categories (psychotic, mood, anxiety, and eating disorders), along with specific most-frequent diagnoses in each category (schizophrenia, major depression, panic disorders, and bulimia nervosa, respectively). Past use and non-drug-use groups were similar in their generally reliable reporting of current and past psychiatric disorders. However, current mood and psychotic disorders were less reliably diagnosed in the group with current substance use disorders.