Reliability of the WAIS-R for a mixed sample of psychiatric and neurological patients was determined. Split-half reliabilities and standard errors of measurement were computed for all subtests (exept Digit Span and Digit Symbol) and the Verbal, Performance, and Full Scale IQs. Only the Arithmetic subtest had a reliability estimate which was significantly lower than that of the 35-44-yr.-old standardization group. All other reliabilities and standard errors of measurement approximated those reported for the normative group. It was concluded that the WAIS-R is a reliable instrument for evaluating patients in a clinical setting.