THERE ARE FOUR MAIN EVALUATION PERSPECTIVES: SPONSOR (FOR EXAMPLE, A STATE DIVISION OF MENTAL HEALTH), PROVIDER (A FEDERALLY FUNDED CHMC), THE PROGRAM'S CONSUMERS (EX-PSYCHIATRIC PATIENTS), AND COMMUNITY RESIDENTS (THE RESIDENTS IN THE NEIGHBORHOOD SURROUNDING THE PROGRAM). AN EXAMPLE OF A TRANSITIONAL APARTMENT PROGRAM FOR FORMER STATE PSYCHIATRIC PATIENTS RETURNING TO THE COMMUNITY ILLUSTRATES THESE PERSPECTIVES. AGENCY EVALUATORS WORKING UNDER MULTIPLE FUNDING AND ACCOUNTABILITY REQUIREMENTS USUALLY EXPERIENCE INCONSISTENT ROLE DEMANDS THAT INTERFERE WITH THE EVALUATION PROCESS. ALSO, EVALUATORS CAN NOT IDENTIFY THE LIKELY OUTCOMES OF PROGRAM ACTIVITIES AND PROGRAM MANAGERS' ACTION DECISIONS IN ACTUAL PRACTICE AS QUICKLY AS IN EVALUATION PROCESSES. EVALUATION ACTIVITIES FOR THE CHMC COVER MIS (MANAGEMENT INFORMATION SYSTEM), NEEDS ASSESSMENT, UTILIZATION, IMPACT, CONSUMER EVALUATION, TREATMENT OUTCOME, STAFF ATTITUDE, PLANNING, AND COST EFFECTIVENESS. THE IMPACT OF EVALUATION ON POLICY IS ENHANCED WHEN THE EVALUATION REPORTS CONCENTRATE ON DECISIONS RATHER THAN CONCLUSIONS, ON APPLICATIONS RATHER THAN IMPLICATIONS, AND UTILIZATION RATHER THAN REPLICATION. THE EVALUATION'S IMPACT ON POLICY IS FURTHER STRENGTHENED WHEN AN INDIVIDUAL TAKES DIRECT PERSONAL RESPONSIBILITY FOR GETTING THE INFORMATION TO THE RIGHT PEOPLE. OTHER FACTORS ARE THE POLITICAL CONTEXT IN WHICH THE EVALUATION IS USED, PROACTIVE RATHER THAN REACTIVE TECHNIQUES IN EXPLORING EVALUATION OUTCOMES AND DECISION ALTERNATIVES, AND ADOPTION OF A MORE REALISTIC EXPECTATION FOR THE IMPACT OF THE EVALUATION. TABLES AND REFERENCES ARE PROVIDED. (PRG)