FROM 1971 TO 1975, THE URBAN INSTITUTE WAS INVOLVED IN EVALUATING A TEAM POLICING PROGRAM IN CINCINNATI, OHIO. THE EXPERIENCE IN CINCINNATI SERVED TO PROVIDE A MEASURE OF SUPPORT TO THE ARGUMENT AGAINST THE USE OF BROADLY DEFINED VICTIMIZATION SURVEYS FOR THE PURPOSE OF PROGRAM EVALUATION. ONE OF THE PRIMARY GOALS OF THE TEAM POLICING PROGRAM WAS THE REDUCTION OF CRIME. SPECIFIC ATTENTION IS GIVEN TO THE SYSTEMATIC ATTEMPT TO RELATE VICTIMIZATION SURVEYS TO MEASURES OF CRIME REPORTED BY THE POLICE AND TO THE USE OF VICTIMIZATION REPORTS TO MEASURE THE PROGRAM'S IMPACT ON CRIME. THE NATURE OF THE TEAM POLICING CONCEPT AND THE DECISION TO USE VICTIMIZATION STUDIES AS EVALUATION TOOLS ARE DISCUSSED. THE DECISION WAS PARTLY BASED ON THE FACT THAT MANY CRIMES GO UNREPORTED; THUS, ANY PROGRAM THAT ATTEMPTS TO SIMULTANEOUSLY REDUCE CRIME AND IMPROVE POLICE-COMMUNITY RELATIONS MAY FIND THAT A REDUCTION IN VICTIMIZATION IS MASKED BY A CORRESPONDING INCREASE IN THE PROPORTION OF CRIME REPORTED. THE RELATIVE SCOPE OF THE MEASURES OF CRIME DEVELOPED BY MEANS OF VICTIMIZATION SURVEYS IS EXAMINED, AND SPECIFIC CRIMES, SUCH AS RAPE AND BURGLARY, ARE DISCUSSED IN THIS REGARD; TABULAR COMPARISONS OF VARIOUS SURVEY RESULTS REGARDING RATES OF RAPE, ROBBERY, AGGRAVATED ASSAULT, BURGLARY, AND AUTO THEFT, ARE PROVIDED. THE QUESTION OF STABILITY OF VICTIMIZATION ESTIMATES IN CINCINNATI IS DISCUSSED. THE SOURCES OF ERROR THAT RESULT IN THE DIFFERENCES IN MEASUREMENTS OF CRIME RATES ARE CONSIDERED. IT IS HYPOTHESIZED THAT THE UNIFORM CRIME REPORT DEFINITIONS OF CRIME, AS OPERATIONALIZED IN THE SURVEY INSTRUMENTS THEMSELVES, CONTRIBUTE SUBSTANTIALLY TO THE DIFFERENCES. AN EXAMPLE OF A STABLE VICTIMIZATION SURVEY IS PRESENTED INCLUDING TABULAR DATA; BASICALLY, IT REVOLVES AROUND A COMMERCIAL VICTIMIZATION SURVEY THAT EMPLOYS A MORE NARROWLY DEFINED DESCRIPTION OF CRIME. IT IS CONCLUDED THAT IN THE CASE OF BROADLY DEFINED VICTIMIZATION SURVEYS, THE VARIANCE RESULTING FROM THE DIFFERENT ADMINISTRATIONS OF THE SAME SURVEY INSTRUMENT MAY BE QUITE LARGE, BUT THAT THE CAREFUL SELECTION AND DEFINITION OF INCIDENTS AND SURVEY POPULATIONS APPEAR TO ALLOW STABLE ESTIMATES FROM WHICH VALID INFERENCES CONCERNING THE EFFECT OF CRIME-REDUCTION PROGRAMS CAN BE MADE. REFERENCE NOTES ARE PROVIDED. (KBL)