A MODEL IS PROPOSED TO EXPLAIN HOW POTENTIAL CLIENTS SELECT AGENCIES, HOW STIGMA BECOME ATTACHED TO FORMER CLIENTS OF CERTAIN AGENCIES, AND WHY IT IS DIFFICULT TO GAIN ACCEPTANCE, BOTH FROM AGENCIES AND FROM THE PUBLIC, OF NEW ROLES FOR AGENCIES. IN THE MODEL, EACH AGENCY HAS SEVERAL MANDATES--SETS OF PRECONDITIONS AND GOALS DEFINING SITUATIONS IN WHICH IT IS APPROPRIATE TO CALL ON THE AGENCY. HYPOTHESES DERIVED FROM THE MODEL ARE TESTED IN A STUDY OF POLICE AND MENTAL HEALTH AGENCY MANDATES VIS-A-VIS FAMILY DISPUTES. THREE 32-MEMBER GROUPS (POLICE OFFICERS, EMPLOYEES OF A COMMUNITY MENTAL HEALTH CENTER, AND MEMBERS OF THE GENERAL PUBLIC) REACTED TO A SERIES OF FAMILY DISPUTE CASE DESCRIPTIONS, RATING THE KEY FEATURES OF EACH SITUATION AND INDICATING THE PROBABILITY THAT POLICE OR PSYCHIATRIC INTERVENTION WOULD OCCUR. THEIR RESPONSES PROVIDE PARTIAL SUPPORT FOR THE MODEL. DIFFERENT FEATURES ARE ASSOCIATED WITH SITUATIONS IN WHICH POLICE INTERVENTION AND MENTAL HEALTH INTERVENTION IS EXPECTED. HOWEVER, THERE IS ONLY LIMITED SUPPORT FOR THE NOTIONS THAT THESE FEATURES ARE ORGANIZED IN THOUGHT AND PERCEPTION INTO DISTINCT MANDATES FOR EACH AGENCY (ORDER MAINTENANCE AND LAW ENFORCEMENT FOR POLICE, CONSTRUCTIVE CHANGE AND MANAGEMENT OF THE INSANE FOR MENTAL HEALTH AGENCIES) AND THAT THESE DISTINCT MANDATES INFLUENCE INFERENCES ABOUT FAMILY DISPUTE SITUATIONS IN WHICH A PARTICULAR AGENCY IS KNOWN TO BE INVOLVED. VERY LIMITED EVIDENCE IS FOUND TO SUPPORT THE HYPOTHESIS THAT INFORMATION ABOUT AN INTERVENING AGENT'S REACTION TO A SITUATION HAS AN EFFECT ON INFERENCES ABOUT THE SITUATION. THERE ARE CERTAIN DIFFERENCES AMONG THE THREE SUBJECT GROUPS, BUT GENERALLY THEY ARE INFLUENCED BY THE SAME FEATURES IN APPROXIMATELY THE SAME WAYS IN MAKING INFERENCES ABOUT FAMILY DISPUTE SITUATIONS. SUGGESTIONS FOR REFINING THE PROPOSED MODEL ARE PRESENTED, AND PRACTICAL IMPLICATIONS OF THE FINDINGS FOR COMMUNITY SERVICE AGENCIES ARE DISCUSSED. SUPPORTING DATA AND A LIST OF REFERENCES ARE PROVIDED.