BY TRACING THE CRIMINAL CAREERS OF OFFENDERS FROM THE TIME OF THEIR FIRST ARREST UNTIL THEIR CAREER IN CRIME ENDS, THE MODEL STRUCTURALLY MIMICS A REPRESENTATIVE CRIMINAL JUSTICE SYSTEM WITH A RECIDIVISM COMPONENT TO ALLOW FOR AN OFFENDER'S REARREST. THE GENERALIZED NETWORK SIMULATOR IS THE VEHICLE FOR IMPLEMENTATION SO THAT QUEUEING, RESOURCE ALLOCATION, AND COST ISSUES CAN ALSO BE CONSIDERED. A QUASI-EXPERIMENTAL DESIGN EXAMINING PLEA BARGAINING ALTERNATIVES DEMONSTRATES THE MODEL'S UTILITY IN EVALUATING CRIMINAL JUSTICE SYSTEM PERFORMANCE. A LINEAR CAREER CRIMINAL SOCIAL COST FUNCTION IS POSTULATED WHICH COMBINES MEASURES OF RECIDIVISM AND CRIMINAL JUSTICE SYSTEM COST. BY MINIMIZING THE CAREER CRIMINAL SOCIAL COST ACCORDING TO DIFFERENT SETS OF IMPORTANCE WEIGHTS FOR MALE AND FEMALE OFFENDERS, THIS RESEARCH INDICATES THAT HAVING THE PROSECUTOR MAKE PLEA BARGAINING DECISIONS BASED ON THE NUMBER OF OFFENSES IN AN OFFENDER'S ARREST RECORD RESULTS IN A SIGNIFICANT SAVINGS OVER SCENARIOS IN WHICH THIS INFORMATION IS NOT AVAILABLE OR IN WHICH AN A PRIORI MEASURE OF THE SEVERITY OF A CRIME CATEGORY IS USED AS A BASIS FOR THE PROSECUTOR'S DECISION. THIS ANALYSIS ALSO SHOWS THAT EACH POLICY ALTERNATIVE AFFECTS DIFFERENT SUBPOPULATIONS OF OFFENDERS IN DIFFERENT MANNERS; THUS, THE CHOICE OF THE IMPORTANCE WEIGHTS NOT ONLY DETERMINES THE OPTIMAL POLICY, BUT IT RAISES THE ISSUE OF EQUITY IN DEALING WITH OFFENDER SUBPOPULATIONS IN CRIMINAL JUSTICE DECISIONMAKING. TABULAR AND GRAPHIC DATA ARE PROVIDED, ALONG WITH A BIBLIOGRAPHY. APPENDIXES CONTAIN AN INPUT DESCRIPTION, SAMPLES OF INPUT AND OUTPUT DATA, AND CRIMINAL JUSTICE SYSTEM SUBROUTINE LISTINGS. (AUTHOR ABSTRACT MODIFIED--RCB)