With proposed criteria for DSM-IV substance dependence imminent, an evaluation of the impact of changes from DSM-III to DSM-III-R would be informative. Recent admissions to St. Louis drug treatment centers were interviewed with the DIS-III-R, which covers criteria from both systems. Kappa values for system agreement, diagnostic overlap and percent positive agreement are reported by substance. The DSM-III-R system cast a wider net for dependence than DSM-III for alcohol, tobacco and amphetamines. Neither system predominated for cannabis, opioids and barbiturates/sedatives/hypnotics. Reasons for differences and implications of findings are discussed.