This report presents results of a field trial of Substance Use Disorders as defined by DSM-III-R, DSM-IV (proposed) and ICD-10. Diagnoses based on the three systems were derived from interviews using the Composite International Diagnostic Interview (CIDI) in a heterogeneous sample of 521 adults drawn from clinical and community settings. Two issues are addressed: (1) cross system agreement; and (2) syndrome coherence of proposed criterion sets for Substance Dependence in each of the three systems. Findings were as follows: (1) Cross system agreement for Dependence was generally high, especially between DSM-III-R and DSM-IV. (2) Cross system agreement was lower for DSM-III-R and DSM-IV Abuse and very low for DSM-IV Abuse and ICD-10 Harmful Use. (3) Agreement varied across drug categories with lowest DSM-III-R/DSM-IV agreement for alcohol abuse and DSM-IV/ICD-10 agreement for marijuana use disorders. (4) Overall prevalence differed for the three systems with DSM-IV yielding highest rates followed by DSM-III-R and ICD-10 in that order. (5) Factor analysis of Dependence criteria showed high loadings of all items on a single factor across the three diagnostic systems and for all categories of drugs. Implications for validity of the dependence syndrome construct and for revisions in DSM-IV are discussed.