UNTIL THIS STUDY, NO ONE HAD DIRECTLY TESTED THE RELATIONSHIP BETWEEN CITIZEN EVALUATION OF POLICE PERFORMANCE AND UNREPORTED CRIME RATE, USING CITIES AS THE UNITS OF ANALYSIS. THE UNREPORTED CRIME RATE IS HYPOTHESIZED TO BE LOWER IN CITIES WITH HIGH POLICE EVALUATION SCORES. IN ORDER TO TEST THE HYPOTHESIS, DATA WERE USED FROM 13 AMERICAN CITIES WITH AVAILABLE POLICE EVALUATION MEASURES. RIVAL HYPOTHESES WERE TAKEN INTO ACCOUNT BY INTRODUCING A NUMBER OF CONTROL VARIABLES REPRESENTING POLICE SYSTEMS ACTIVITIES AND POPULATION CHARACTERISTICS. ESTIMATED 1975 POPULATION, 1970 POPULATION DENSITY, AND 1970 PERCENTAGE OF NONWHITE POPULATION ARE FROM THE U.S. BUREAU OF THE CENSUS. UNREPORTED CRIME RATES WERE ESTIMATED BY SUBTRACTING OFFICIAL CRIME RATES COMPILED BY THE FBI FROM THE UNOFFICIAL VICTIMIZATION RATES REPORTED BY THE NATIONAL CRIME SURVEY. POLICE EVALUATION WAS MEASURED BY CALCULATING A MEAN RESPONSE SCORE FOR EACH CITY FROM A FOUR-POINT ATTITUDINAL SCALE. POLICE EVALUATION WAS FOUND NOT TO BE A STATISTICALLY SIGNIFICANT DETERMINATE OF UNREPORTED CRIME FOR THE CITIES STUDIED. THE PERCENTAGE OF NONWHITES AND POLICE SYSTEMS ACTIVITIES ARE STRONGLY RELATED TO THE UNREPORTED CRIME RATES, AS IS POPULATION DENSITY, WITH LESS CRIME UNREPORTED IN CITIES WITH HIGH PERCENTAGES OF MINORITIES AND A RELATIVELY HIGH NUMBER OF PATROLMEN PER CAPITA. NOTES AND REFERENCES ARE PROVIDED. (RCB)