Research evaluating the reliability of the Structured Clinical Interview for DSM-III-R (SCID) is reviewed. Reliability procedures and studies are examined. Several versions of the SCID are covered, including the SCID-I (axis I disorders), SCID-II (axis II disorders), SCID-Positive and Negative Syndrome Scale (SCID-PANSS; functional-dimensional assessment for psychotic disorders), and SCID-Upjohn Version (panic disorder). The SCID has been found to yield highly reliable diagnoses for most axis I and axis II disorders. Suggestions for future research on the SCID are offered, particularly with respect to (1) the lack of studies in which SCID diagnoses are compared with diagnoses from unstructured interviews or other structured-interview formats, and (2) the need for a more natural evaluation of this instrument. Also, the importance of establishing norms and obtaining reliability data for underserved clinical populations is discussed.