THE DEVELOPMENT OF RATIONAL SENTENCING POLICIES MUST OCCUR THROUGH A SYSTEMATIC ASSESSMENT OF SENTENCING GOALS AND EMPIRICAL ANALYSIS OF DATA RELEVANT TO THESE GOALS. IN SUPPORT OF THESE GENERAL PROPOSITIONS AND IN ANTICIPATION OF LEGISLATION REQUIRING RESEARCH ON SENTENCING POLICIES, THE FEDERAL JUSTICE RESEARCH PROGRAM OF THE DEPARTMENT OF JUSTICE CONTRACTED WITH THE INSTITUTE FOR LAW AND SOCIAL RESEARCH TO CONDUCT RESEARCH FOR USE IN FORMULATING FEDERAL SENTENCING GUIDELINES. PREVIOUS RESEARCH RELEVANT TO SENTENCING HAS REVEALED, ALTHOUGH IT CANNOT BE PROVED SCIENTIFICALLY, THAT SENTENCES ACTUALLY IMPOSED CONTAIN UNWARRANTED VARIATION, SOME LACK OF EVENHANDEDNESS IS INFERABLE FROM AVAILABLE DATA. LITTLE OR NO REDUCTION IN RECIDIVISM OCCURS, REGARDLESS OF THE CORRECTIONAL STRATEGIES USED. STUDIES DEALING WITH THE EFFECT OF SENTENCING ON REHABILITATION AND DETERRENCE CURRENTLY HAVE SERIOUS LIMITATIONS, BUT IMPROVEMENTS IN ESTIMATES CAN BE EXPECTED. THE STUDY PLAN UNDERWAY CONSISTS OF THREE MAJOR COMPONENTS: (1) AN ANALYSIS OF SENTENCING SEVERITY, (2) AN ANALYSIS OF THE EFFECTS OF SENTENCING DECISIONS ON CRIME CONTROL AND INDIVIDUAL BEHAVIOR, AND (3) A SURVEY OF CRIMINAL JUSTICE PERSONNEL AND THE GENERAL PUBLIC TO PROVIDE INFORMATION ABOUT THE ACCEPTABILITY AND EFFECTIVENESS OF ALTERNATIVE SENTENCING POLICIES. A UTILITARIAN MODEL CAN BE USED TO PRODUCE SENTENCING GUIDELINES FROM THE RESEARCH ESTIMATES. THE MODEL PREFERRED PROVIDES AN EXPLICIT FRAMEWORK FOR INTEGRATING THE CRIME CONTROL EFFECTS OF SENTENCING, THE COSTS OF CONTROLLING CRIME, AND THE PRIVATELY INCURRED COSTS OF CRIME. BY INCORPORATING ESTIMATES OF EACH OF THESE EFFECTS AND COSTS INTO THIS MODEL, AN ESTIMATE OF THE SOCIALLY OPTIMAL SENTENCE UNDER THE UTILITARIAN LOGIC CAN BE CALCULATED. FOOTNOTES ARE PROVIDED. (RCB)