THE KEY TO PPBS ANALYSIS IS SELECTING THE CORRECT OUTPUTS TO BE MEASURED, A SITUATION MADE MORE DIFFICULT BY THE LACK OF CLEAR RELATIONSHIP BETWEEN INPUTS AND OUTPUTS IN POLICE WORK. FOR THIS REASON, CONTROLLED EXPERIMENTS OFFER MORE RELIABLE RESULTS THAN MULTIVARIATE ANALYSES RUN ON AN EXISTING SITUATION. ALSO, THE RESULTS DESIRED MUST BE CAREFULLY DEFINED; IS THE PROGRAM TO REDUCE CRIME RATES, INCREASE CITIZENS' FEELINGS OF SECURITY, OR REDUCE TRAFFIC VIOLATIONS? THREE CONTROLLED EXPERIMENTS USING WELL-DEFINED CONDITIONS AND PROVIDING USEFUL ANSWERS ARE THE KANSAS CITY PREVENTIVE PATROL EXPERIMENT, THE 20TH PRECINCT EXPERIMENT IN NEW YORK CITY, AND THE LOS ANGELES TRAFFIC EXPERIMENT. IN THE KANSAS CITY STUDY, TEST PRECINCTS WERE DIVIDED INTO GROUPS OF THREE MATCHED FOR SOCIOECONMIC CHARACTERISTICS, POPULATION, AND TYPE OF CRIME PROBLEM, ROUTINE BEAT SERVICE WAS MAINTAINED IN THE CONTROL DISTRICTS. IN ONE OF THE TEST DISTRICTS, POLICE CARS ENTERED THE BEATS ONLY IN RESPONSE TO SPECIFIC CALLS. IN THE OTHER TEST DISTRICT, HIGH-VISIBILITY PATROLS WERE ESTABLISHED WITH TWO TO THREE TIMES THE CARS ORDINARILY USED. FINDINGS WERE THAT CRIME RATES AND CITIZEN PERCEPTION OF CRIME WERE SIMILAR IN BOTH THE NO-PATROL AND THE HIGHLY PATROLLED DISTRICTS. IT WAS CONCLUDED THAT TRADITIONAL POLICE PATROL DOES LITTLE TO REDUCE CRIME. IN CONTRAST, IN THE NEW YORK EXPERIMENT A 40 PERCENT INCREASE IN MANPOWER LED TO A SIGNIFICANT REDUCTION IN 'OUTSIDE' CRIMES OF ROBBERY, LARCENY, AUTO THEFT, TOTAL FELONIES, AND MISDEMEANORS. 'OUTSIDE' CRIMES ARE DEFINED AS THOSE THE PATROLMAN COULD SEE WITHOUT LEAVING HIS CAR. 'INSIDE' CRIMES SHOWED LITTLE DECREASE. MATCHED DISTRICTS WERE USED IN THE LOS ANGELES EXPERIMENT; THE NUMBER OF TRAFFIC ACCIDENTS WITH INJURIES WERE THE MEASURED OUTPUT SINCE THESE ARE MORE LIKELY TO BE REPORTED THAN PROPERTY-DAMAGE-ONLY ACCIDENTS. THIS STUDY FOUND THAT ISSUING CITATIONS WAS LESS EFFECTIVE THAN ISSUING WARNINGS BECAUSE CITATIONS WERE USUALLY LIMITED TO CASES WHICH COULD HOLD UP IN COURT; THE MARGINAL VIOLATIONS WERE THE MAJOR CAUSES OF ACCIDENTS. INCREASED MANPOWER ALSO DECREASED ACCIDENTS. THIS EXPERIMENTAL FINDING CAN ENABLE DEPARTMENT MANAGERS TO EMPHASIZE WARNINGS RATHER THAN CITATIONS AND JUSTIFY INCREASED PATROL TO REDUCE ACCIDENTS. USE OF PPBS TO IMPROVE OR MODIFY PRESENT PROGRAMS CAN BE MORE EFFECTIVE THAN USING IT ONLY TO START NEW PROGRAMS OR TO MAKE DECISIONS ABOUT ABANDONING AN EFFORT. REFERENCES ARE INCLUDED.