A total of 2,061 client files were abstracted for the analyses. Variables for which data were collected were demographics, place of referral, problems indicated at the time of interview, prior mental health treatment (including alcoholism treatment), Mortimer-Filkins score, blood alcohol concentration at the time of the arrest, and blood chemistry test scores that were completed shortly before or after the interview. Analyses include comparisons between percentages, measures of association, and differences between means for those not-recommended and those recommended for treatment. A discriminant function analysis identified the variables that combined in a linear model to separate the two groups as distinctly as possible. Based on clinical evaluations at the alcoholism treatment program, 1,470 were recommended for treatment, and 591 were not recommended. The Mortimer- Filkins score and the blood alcohol concentration at the time of arrest were the two best discriminating variables in the decision to recommend treatment. When these scores were high, additional information was not needed; however, when these scores were in the low or midranges, other variables became crucial to the decisionmaking process. Further investigation of variables not available in these analyses is required to determine the decisionmaking process for clients with low or midrange Mortimer-Filkins and blood- alcohol-concentration scores. The Mortimer-Filkins test identifies problem areas regarding family, stress, employment, finances, depression, nervousness, drinking, feelings toward oneself, and ability to cope with life. 2 tables, 1 figure, and 23 references