THE STUDY SEEKS TO EXPLAIN WHY DECISONMAKERS OFTEN FAIL TO USE THE RESULTS OF FORMAL EVALUATION STUDIES IN ARRIVING AT POLICY, PROGRAM, AND CLINICAL DECISIONS. A QUESTIONNAIRE WAS SENT TO 377 SUBJECTS (190 RESPONSES) ASKING THEM TO RATE EACH OF NINE AGENCIES IN A DRUG TREATMENT SYSTEM AND TO ESTIMATE THE INFLUENCE OF SEVERAL INFORMATION SOURCES AND DECISION CRITERIA ON THEIR RATINGS. RESPONDENTS INCLUDE AGENCY STAFF, PLANNERS, COURT PERSONNEL, AND PROBATION AND PAROLE PERSONNEL. THE FINDINGS SHOW THAT DECISIONMAKERS IN DIFFERENT OCCUPATIONS TEND TO USE SIMILAR INFORMATION SOURCES (PRIMARILY INFORMAL VERBAL EXCHANGES) BUT DIFFERENT EVALUATIVE CRITERIA TO JUDGE TREATMENT PROGRAMS. STAFF MEMBERS OF HUMAN SERVICE ORGANIZATIONS CONSIDER STAFF COMPETENCE TO BE THE MOST IMPORTANT CRITERION, PLANNERS CITE MANAGEMENT EFFICIENCY AS MOST IMPORTANT, AND JUDICIAL AND PROBATION PERSONNEL BOTH SELECT CLIENT IMPACT AS THE MOST IMPORTANT CRITERION. THE COST OF AGENCY OPERATIONS IS AMONG THE LEAST INFLUENTIAL CRITERIA FOR ALL OCCUPATIONS. THE FINDINGS IMPLY THAT EVALUATORS SHOULD CONSIDER THE INFORMATION NEEDS OF DECISIONMAKERS IN DESIGNING EVALUATIONS, INCLUDE DIRECT PERSONAL CONTACTS AND ORAL PRESENTATIONS IN REPORTING EVALUATION FINDINGS, AND RECOGNIZE THE POSSIBILITY THAT EVALUATION OUTCOME DATA WILL BE OF INTEREST TO PRACTITIONERS YET WILL HAVE LITTLE INFLUENCE ON THEIR DECISIONS. THE FINDINGS ALSO HAVE IMPLICATIONS FOR STRATEGIES IN THE SELECTION OF EVALUATION ADVISORY COMMITTEE MEMBERS. SUPPORTING DATA AND A LIST OF REFERENCES ARE INCLUDED. (LKM)