State substance abuse management information systems increasingly are becoming important tools for research, program management, and policy formulation at federal and state levels. These systems are currently undergoing radical expansion, leading to the creation of statewide performance and outcome monitoring systems for publicly-funded substance abuse treatment. This expansion makes imperative increased knowledge of the psychometric properties of the data in these systems. This study develops a method for examining the reliability of such data and applies it to the Massachusetts Substance Abuse Management Information System (SAMIS). Cohen's κ, intraclass correlations, and the techniques of Heise (1969) are used to assess the reliability of different types of variables. Results show that key variables on the SAMIS Admission Form exhibit moderate to high reliability, supporting the use of this data for aggregate analyses. At the same time, caution should be used in making judgments about individual patients. © 2002 Wiley Periodicals, Inc. J Clin Psychol 58: 1057–1069, 2002.