THE RESEARCH WAS BASED ON THE PREMISE THAT VARIATION IN LAW ENFORCEMENT ACTIVITY CANNOT BE EXPLAINED SOLELY BY DIFFERENT LEVELS OF CRIMINAL BEHAVIOR AMONG COMMUNITIES. EXPLAINING ARREST VARIATION ENTAILS IDENTIFYING THE FACTORS WHICH INFLUENCE THE MANNER IN WHICH POLICE DISCRETION IS EXERCISED. A NUMBER OF HYPOTHESES WERE OFFERED TO EXPLAIN WHY SOME POLICE AGENCIES INVOKE THE CRIMINAL PROCESS MORE FREQUENTLY THAN OTHERS. POSSIBLE INFLUENCES ON ARREST VARIATION WERE EXAMINED WHICH, IN ADDITION TO THE CRIME ENVIRONMENT, INCLUDED A CITY'S SOCIOECONOMIC-CULTURAL ENVIRONMENT, THE LOCAL POLITICAL CLIMATE, LOCAL GOVERNMENT INSTITUTIONS AND PROCESSES, POLICE SYSTEM VARIABLES, AND CERTAIN EXTRACOMMUNITY FACTORS. ARRESTS WERE GROUPED INTO CATEGORIES, AS IT WAS HYPOTHESIZED THAT ARRESTS FOR PART I OFFENSES OR INDEX CRIMES (SERIOUS CRIMES) WOULD BE MORE SENSITIVE TO THE CRIME ENVIRONMENT THAN THOSE FOR PART II CRIMES ON THE ASSUMPTION THAT THE AMOUNT OF DISCRETION IS INVERSELY RELATED TO THE SERIOUSNESS OF THE CRIME. PART I ARRESTS WERE FURTHER DIVIDED INTO PROPERTY AND NONPROPERTY CRIMES, AND PART II ARRESTS WERE DIVIDED INTO CATEGORIES OF VICTIMLESS CRIMES AND ORDER MAINTENANCE OFFENSES ON THE PREMISE THAT ARREST POLICIES MAY BE RESPONSIVE TO DIFFERENT INFLUENCES DEPENDING ON THE TYPE OF VIOLATION INVOLVED. SIMPLE AND PARTIAL CORRELATION ANALYSIS WAS CONDUCTED TO SELECT FROM A LARGE NUMBER OF INDEPENDENT VARIABLES A SMALLER SET OF INDICATORS REPRESENTATIVE OF THE VARIOUS COMPONENTS IN THE DESIGN. SUBSEQUENT MULTIVARIATE ANALYSIS INDICATED THAT THE INDEPENDENT VARIABLES DID NOT BEHAVE CONSISTENTLY WITHIN CATEGORIES OF ARREST, SUGGESTING THAT WHILE SOME TYPES OF CRIMES MAY SHARE CERTAIN CHARACTERISTICS, THESE SHARED CHARACTERISTICS ARE NOT NECESSARILY A BASIS FOR ASSUMING THAT ARRESTS FOR THESE CRIMES WILL BE SIMILARLY INFLUENCED BY A COMMON SET OF FACTORS. ARRESTS FOR PART I OFFENSES INVOLVING AN ELEMENT OF VIOLENCE WERE THE MOST RESPONSIVE TO THE CRIME ENVIRONMENT. THE SOCIOECONOMIC, POLICE DEPARTMENT, AND POLITICAL CLIMATE VARIABLES WERE NOT UNIMPORTANT IN PREDICTING ARREST PATTERNS. ONE MEASURE OF POLICE DEPARTMENT STYLE -- TASK ORIENTATION -WAS POSITIVELY RELATED TO ALL CATEGORIES OF ARREST. A MEASURE OF A COMMUNITY'S PUBLIC-RELATING ORIENTATION WAS POSITIVELY RELATED TO ALL CATEGORIES OF ORDER MAINTENANCE OFFENSES, SUGGESTING THE IMPORTANCE OF CITIZEN INPUT IN THE EXERCISE OF POLICE DISCRETION. THE DATA ON THE IMPACTS OF POLICE BEHAVIOR SUGGESTED THAT VARIOUS PATTERNS OF ARREST FOR LESS SERIOUS OFFENSES SUCH AS VICTIMLESS CRIMES DO NOT IMPAIR THE ABILITY OF THE POLICE TO APPREHEND INDIVIDUALS WHO HAVE COMMITTED MORE SERIOUS CRIMES. ON THE OTHER HAND, GREATER ARREST ACTIVITY FOR LESS SERIOUS CRIME DOES NOT SEEM TO DEFLATE FUTURE CRIME RATES. THEORETICAL FRAMEWORK, RESEARCH DESIGN, AND DATA ANALYSIS ARE DESCRIBED. SUPPORTING DOCUMENTS AND STATISTICS ARE APPENDED, AND REFERENCE AND TABLES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--DS)