FOLLOWING A REVIEW OF STUDIES DEALING WITH THE PROBLEM OF SENTENCING DISPARITY, THE METHODOLOGY AND FINDINGS ARE PRESENTED OF A STUDY OF THE SENTENCES GIVEN TO 1,665 OFFENDERS WHO, AFTER ARREST IN 1974, WERE CONVICTED OF FELONIES IN THE SUPERIOR COURT OF THE DISTRICT OF COLUMBIA. DATA ON FACTORS RELATED TO SENTENCING WERE DRAWN FROM THE PROSECUTOR'S MANAGEMENT INFORMATION SYSTEM, BAIL AGENCY FILES, COURT FILES, AND THE BIOGRAPHICAL SKETCHES OF JUDGES AND PROSECUTORS. TABULAR ANALYSIS WAS DONE OF THE DISTRIBUTION OF SENTENCES UNDER A VARIETY OF CONDITIONS, AND MULTIVARIATE ANALYSIS OF THE DETERMINANTS OF THE SENTENCING DECISION WAS CONDUCTED. IT WAS FOUND THAT THE MOST IMPORTANT DETERMINANT OF THE DECISION ABOUT WHETHER TO INCARCERATE THE OFFENDER WAS HIS PRIOR CRIMINAL RECORD, WHILE THE NATURE OF THE FELONY OFFENSE WAS MOST INFLUENTIAL IN DETERMINING SENTENCE LENGTH. IT WAS ALSO DISCOVERED THAT CONSIDERABLE SENTENCING VARIATION EXISTED BETWEEN JUDGES, WITH SUCH DIFFERENCES NOT BEING ATTRIBUTABLE TO TYPE OF CASELOAD OR TYPE OF DEFENDANT ADJUDICATED. THE LACK OF AN OVERALL PATTERN OF RATIONAL SENTENCE DECISIONMAKING PREVENTED THE CONSTRUCTION OF A MODEL FOR SENTENCING DECISIONS. IT IS PROPOSED THAT SENTENCING VARIATION BE REDUCED BY SPECIFYING WITHIN A GIVEN COURT SYSTEM A SET OF NORMS BASED ON THE PRIOR SENTENCING PRACTICES OF THAT COURT. THIS WOULD NOT MEAN, HOWEVER, THAT JUDGES WOULD BE BOUND BY SENTENCING PRACTICES OF THE PAST, BUT A FRAME OF REFERENCE WOULD BE ESTABLISHED FOR JUDICIAL REVIEW IN SETTING STANDARDIZED POLICY FOR ALL THE JUDGES SENTENCING IN A PARTICULAR COURT. THE PROCEDURES AND FORMS REQUIRED FOR SUCH A SYSTEM ARE DESCRIBED. THE APPENDIXES INCLUDE A SELECTED BIBLIOGRAPHY OF EMPIRICAL STUDIES PERTAINING TO SENTENCING DISPARITY, VARIABLES USED IN ANALYSIS, AND PROBIT ANALYSIS. THE DATA USED IN THE STUDY ARE INCLUDED. (RCB)