THE FIRST MODEL OF CRIMINAL BEHAVIOR, ERLICH'S MODEL, CONSIDERS HOW MUCH AND WHAT TYPE OF CRIMES AN INDIVIDUAL WILL COMMIT. THE SECOND MODEL IS BASED ON THEORIES OF THE FACTORS INFLUENCING THE SENTENCING JUDGES' DECISION; THE FACTORS INCLUDE THE TYPE OF CRIME, PREVIOUS CRIMINAL RECORD, DEMOGRAPHIC BACKGROUND, AND BEHAVIOR IN COURT. THE INFLUENCE OF THE RESULTING SENTENCE ON THE OFFENDERS RECIDIVISM PROBABILITY IS ALSO CONSIDERED, AND THESE HYPOTHESES ARE PRESENTED: JUDGES DISCRIMINATE AGAINST PERSONS WHO HAVE PLEA BARGAINED AND BLACKS, AND AN INDIVIDUAL'S CRIME IS RELATED TO EDUCATIONAL OR SKILL LEVEL. THE FIRST MODEL IS TESTED USING NONLINEAR REGRESSION WITH THE PROBABILITY OF A FELON'S RECIDIVISM, THE PROBABILITY OF A FELON'S BEING WHITE, AND THE PROBABILITY OF A FELON'S COMMITTING A SKILLED CRIME AS DEPENDENT VARIABLES. THE RESULTS GENERALLY SUPPORT THE MODEL. FOR EXAMPLE, PERSONS WHO COMMIT MORE SKILLFUL CRIMES WERE BETTER EDUCATED THAN PERSONS WHO COMMITTED LESS SKILLFUL CRIMES. THE SECOND MODEL IS TESTED USING DATA ON FOUR CRIME CATEGORIES INCLUDING ROBBERY, BURGLARY, ASSAULT-MURDER, AND AUTO THEFT. THE EXPECTED JUDICIAL DISCRIMINATION WAS EVIDENT AGAINST NONWHITES, PERSONS WHO PLEA BARGAIN, AND PERSONS WHO INSIST ON HAVING THEIR CASES TRIED. THERE WAS INTERACTION BETWEEN THE PAROLE BOARD AND THE JUDGES, WITH THE PAROLE BOARD SOMETIMES AGREEING AND SOMETIMES DISAGREEING WITH THE JUDGES' DETERMINATION OF OPTIMAL SENTENCES. IT IS CONCLUDED THAT THE RESULTS WERE CONSISTENT WITH SEVERAL WORKS BY ECONOMISTS IN SHOWING THAT ECONOMIC THEORY CAN GENERATE USEFUL HYPOTHESES ABOUT CRIMINAL BEHAVIOR AND ABOUT SOCIETY'S RESPONSE TO CRIME. THE ERLICH MODEL IS ILLUSTRATED IN THE APPENDIX. STUDY DATA TABLES AND BIBLIOGRAPHICAL REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED-DAG)