To obtain data on patient characteristics relevant to treatment outcome, the Minnesota Multiphasic Personality Inventory (MMPI) was administered to 86 men 3-5 days following their admission to a 30-day residential alcoholism treatment program. Cluster analyses yielded three subtypes whose mean MMPI code lypes were as follows: Type I comprised a 4-2-8 code type, with marked additional pathology: Type II comprised a 4-9/9-4 code type; and Type III comprised a 2-4/4-2 code type. The subtypes differed significantly in their rates of treatment completion, with Type II yielding a greater proportion of program dropouts. To investigate the effect of time of MMPI administration on subtype results, 68 subjects who remained in treatment were readministered the MMPI 14-16 days following admission. A comparison of Time 1 and Time 2 typologies highlights the importance of test administration time and provides some explanation for previous discrepant findings.Although large numbers of people seek treatment for substance abuse problems, it has been reported that less than 30% complete their treatment programs (U.S. Department of Health, Education, and Welfare, 1980). Research to identify the factors that predict treatment success has focused on two areas.First, the importance of investigating treatment features that predict treatment outcome is clear, and a body of literature addressing this issue is beginning to emerge. Second, researchers have investigated the patient variables that predict success in treatment programs.Among patient characteristics, background and demographic variables have proven to be powerful predictors of treatment outcome. Demographic variables have been found to account for a large portion of the variance between those who complete treatment programs and those who do not (Hoffman & Jansen, 1973; Hoffman, Jansen, & Wefring, 1972). Many individual treatment programs, however, have tended to attract groups that are somewhat homogeneous demographically.Nondemographic predictions of treatment outcome that use single Minnesota Multiphasic Personality Inventory (MMPI) scales to assess patient characteristics have yielded conflicting results. Huber and Danahy (1975) found that only the Psychopathic Deviate (Pd) scale significantly differentiated completers from noncompleters. Mozdzierz, Macchitelli, Conway, and Krauss (1973) found that noncompleters tended to produce more defensive profiles than did completers, whereas Krasnoff