A LIST OF POLICE ACTIVITIES ALLEGED TO REDUCE CRIME ILLUSTRATES THE DIVERSITY OF TACTICS OPEN TO RESEARCH ON DETERRENT EFFECTS. PROBLEMS IN MEASURING THESE EFFECTS--COUNTING CRIME, DETERMINING THE PROBABILITY OF APPREHENSION, DISPLACEMENT (SHIFTING THE LOCALE OF CRIME RATHER THAN ACTUALLY PREVENTING CRIME), TEMPORAL EFFECTS (CHANGES IN DETERRENT EFFECTS OVER TIME)--ARE REVIEWED. THE FINDINGS AND METHODOLOGICAL DRAWBACKS OF CROSS-SECTIONAL STUDIES (COMPARISONS OF POLICE ACTIVITIES AND CRIME RATES ACROSS JURISDICTIONS), LONGITUDINAL STUDIES (EXAMINATIONS OF CHANGES IN CRIME RATES OVER TIME), THE KANSAS CITY PREVENTIVE PATROL EXPERIMENT, AND STUDIES OF POLICE RESPONSE TIME ARE ASSESSED. IT IS CONCLUDED THAT PROBABILITY OF ARREST APPARENTLY HAS A DETERRENT EFFECT FOR AT LEAST SOME TYPES OF CRIME, BUT THAT MORE INFORMATION IS NEEDED REGARDING WHAT TYPES OF CRIME ARE DETERRED AND TO WHAT EXTENT. IT APPEARS THAT NEARLY ANY SENSIBLY DEPLOYED INCREASE IN POLICE PERSONNEL CAN INCREASE THE NUMBER OF ARRESTS, BUT THE EFFECT OF TOTAL POLICE RESOURCES ON CRIME RATES IS NOT WELL UNDERSTOOD. PROBLEMS OF DATA QUALITY, INTERACTIONS AMONG VARIABLES, AND UNCONTROLLABLE CHANGES IN THE REAL WORLD HAVE HAMPERED RESEARCH AIMED AT DETERMINING HOW FIXED POLICE RESOURCES CAN BEST BE USED TO REDUCE CRIMES. IT DOES APPEAR THAT EXPANDED FOOT PATROL PRODUCES DETECTABLE REDUCTIONS IN CRIME IN TARGET AREAS, ALTHOUGH THE DISPLACEMENT EFFECTS OF SUCH ACTIVITIES AND THE TENACITY OF THE DETERRENT EFFECT OVER TIME ARE NOT KNOWN. NEITHER HAS A DETERMINATION BEEN MADE REGARDING WHETHER IMPROVED RESPONSE TIME CAN INCREASE THE PROBABILITY OF ARREST. THE NEED TO REPLICATE THE KANSAS CITY PATROL EXPERIMENT IN OTHER JURISDICTIONS IS POINTED OUT, AS IS THE NEED FOR A COMPREHENSIVE, LONG-TERM APPROACH TO THE STUDY OF DETERRENCE. SUPPORTING DATA AND A LIST OF REFERENCES ARE INCLUDED. (LKM)