DISTINGUISHING FEATURES OF THE COMMUNITY PSYCHIATRIC SERVICE IN CHICHESTER, ENGLAND, INSTITUTED IN 1958 WERE: (1) NO PATIENT COULD BE ADMITTED WITHOUT FIRST SEEING A PSYCHIATRIST, EITHER AT HOME, A DAY CENTER, OR AN OUTPATIENT CLINIC; (2) CLOSE COLLABORATION WITH LOCAL GENERAL PRACTITIONERS WAS CONSIDERED ESSENTIAL; AND (3) BECAUSE THE SERVICE WAS RUN BY STAFF OF THE PSYCHIATRIC HOSPITAL (GRAYLINGWELL HOSPIAL), ITS EMPHASIS WAS MORE ON CLINICAL THAN ON SOCIAL CARE. AIMS IN EVALUATING COMMUNITY SERVICES IN CHICHESTER WERE DETERMINED BY A COMBINATION OF FACTORS, INCLUDING STATED OBJECTIVES OF THE SERVICE, UNSTATED BUT IMPLICIT GOALS IN PROVIDING NEW FACILITIES FOR TREATMENT IN THE COMMUNITY, GENERAL PURPOSES OF SERVICE EVALUATION SUCH AS REDUCING MORBIDITY AND INCREASING EFFICIENCY, UNEXPECTED AND UNCONSIDERED PROBLEMS CREATED BY THE SERVICE, AND FEASIBILITY OF EACH OBJECTIVE IN TERMS OF RESEARCH DESIGN AND METHOD. AIMS RESULTING FROM CONSIDERATION OF THESE FACTORS WERE TO SEE HOW THE INTRODUCTION OF A COMMUNITY SERVICE AFFECTS WHO IS REFERRED TO A PSYCHIATRIST, TO SEE HOW A COMMUNITY SERVICE AFFECTS WHO IS ADMITTED AND WHO RECEIVES TREATMENT IN THE COMMUNITY, TO DETERMINE THE IMPACT ON THE COMMUNITY OF TREATING PATIENTS OUTSIDE THE HOSPITAL WHO PREVIOUSLY WOULD HAVE BEEN ADMITTED, AND TO ASSESS THE EFFECT OF THE NEW SERVICE ON PATIENT CLINICAL AND SOCIAL OUTCOMES. AFTER FEASIBILITY AND PILOT STUDIES, PLANNING THE EVALUATION, AND THE SELECTION OF A CONTROL SERVICE FOR COMPARISON, DATA WERE COLLECTED AND CRITERIA FOR EVALUATION (INDEXES AND MEASURES) WERE ESTABLISHED. RESULTS INCLUDED EPIDEMIOLOGICAL INDEXES ABOUT THE TARGET POPULATION, MEASURES OF BURDEN ON FAMILIES, OUTCOME MEASURES, AND MEASURES OF SERVICE QUALITY AND ACCEPTABILITY. REFERRAL RATES TO THE COMMUNITY SERVICE WERE HIGHER FOR NEARLY EVERY DEMOGRAPHIC, SOCIAL, AND CLINICAL CATEGORY THAN THE CONTROL SERVICE. IT APPEARED THAT THE NEW SERVICE WAS ESPECIALLY RESPONSIVE TO THE NEEDS OF THE NEGLECTED ELDERLY AND PERSONS WITH SERIOUS MENTAL DISORDERS. ONLY 14 PERCENT OF REFERRALS WERE ADMITTED IN THE MONTH FOLLOWING REFERRAL TO THE COMMUNITY SERVICE, IN CONTRAST TO 52 PERCENT FOR THE HOSPITAL-CENTERED CONTROL SERVICE. ADMISSION RATES WERE 1.0 AND 2.8 PER 1,000, RESPECTIVELY. THIS LOWER ADMISSION RATE IN THE COMMUNITY SERVICE WAS MAINTAINED THROUGHOUT THE 2 YEARS OF EVALUATION. AT THE TIME OF REFERRAL TO BOTH SERVICES, THE EFFECTS OF HAVING A MENTALLY ILL FAMILY MEMBER ON TWO-THIRDS OF THE 410 FAMILIES INVOLVED IN THE EVALUATIVE STUDY WERE MODERATE AND IN ONE-FIFTH OF THE FAMILIES THE EFFECTS WERE SEVERE. THE FINDINGS EMPHASIZE THE IMPORTANCE OF SUPPLEMENTING THE CLINICAL CARE OF PATIENTS TREATED OUTSIDE HOSPITALS WITH A SYSTEMATIC APPRAISAL OF THE NEEDS OF THEIR FAMILIES. SUPPORTING DATA AND NOTES ARE INCLUDED. (DEP)