THE STUDY IS BASED ON PROSECUTOR'S MANAGEMENT INFORMATION SYSTEM (PROMIS) DATA FOR A SAMPLE OF 4,703 ARRESTEES OVER A 56-MONTH PERIOD. ANALYSIS CONFIRMS THAT A SMALL PROPORTION OF ARRESTEES ACCOUNT FOR MANY OF THE ADULT ARRESTS IN THE DISTRICT OF COLUMBIA. CERTAIN CHARACTERISTICS OF DEFENDANTS, READILY IDENTIFIABLE AT INITIAL CASE SCREENING, WERE FOUND TO BE ASSOCIATED WITH FUTURE RECIDIVISM, DEPENDING UPON THE DEFINITION OF RECIDIVISM (REARREST, REPROSECUTION, OR RECONVICTION) AND THE GROUP BEING STUDIED (FELONY OR MISDEMEANOR DEFENDANTS). DEFENDANTS WHOSE CURRENT ARREST WAS FOR FELONY BURGLARY OR MISDEMEANOR BURGLARY WERE VERY LIKELY TO EXPERIENCE FUTURE ARRESTS. THE ASSOCIATION BETWEEN CURRENT OFFENSE TYPE AND THE LIKELIHOOD OF RECIDIVISM SUGGEST THAT CAREER CRIMINAL PROGRAMS SHOULD TARGET THESE OFFENDER TYPES. EFFECTS ON CAREER CRIMINALS' HISTORIES WERE LESS SPECIFIC BECAUSE OF A HIGH INTERCORRELATION AMONG VARIABLES. A NUMBER OF ITEMS--NUMBER OF PREVIOUS ARRESTS, ARRESTS IN THE PAST 5 YEARS, AND NUMBER OF CONVICTIONS--WERE SIGNIFICANT. SEVERAL OTHER VARIABLES HAVE POSSIBLE POLICY SIGNIFICANCE, INCLUDING DEFENDANT'S EMPLOYMENT STATUS, POLICE INDICATION OF DRUG USE, AND LACK OF FAMILIARITY BETWEEN OFFENDER AND VICTIM. PREDICTION ACCURACY WAS HIGHER WHEN MISDEMEANOR ARRESTEES WERE CONSIDERED. A NUMBER OF PROCESSING DECISIONS, TWO DIVERSION PROGRAMS, PATTERNS OF CRIME, AND PROFESSIONALISM OF OFFENDER WERE EXAMINED ALSO IN RELATION TO RECIDIVISM AND TYPES OF CRIMES COMMITTED WHEN REOFFENDING, BUT NO SIGNIFICANT FINDINGS RESULTED. ANALYSIS SUGGESTS SEVERAL AVENUES FOR RESEARCH, SUCH AS THE APPLICABILITY OF THESE FINDINGS TO THE GENERAL CRIMINAL POPULATION, IMPROVEMENT OF RECIDIVISM PREDICTION, AND PROSECUTORIAL DECISIONMAKING BASED ON RECIDIVISM PREDICTION. APPENDIXES CONTAIN STUDY DATA AND DESCRIPTIONS OF STUDY TECHNIQUES. (DAG)