AN EXAMINATION OF THE NATURE AND DEVELOPMENT OF COST-BENEFIT ANALYSIS AND ITS APPLICATIONS IN CRIMINAL JUSTICE CONCLUDES THAT SUCH ANALYSIS IS TYPICALLY NOT AN APPROPRIATE TOOL FOR ASSESSING THE WORTH OF CRIMINAL JUSTICE PROGRAMS. THE COMPLEXITY OF CRIMINAL JUSTICE OBJECTIVES MAKES IT DIFFICULT TO EVALUATE THESE PROGRAMS RIGOROUSLY ENOUGH TO ESTABLISH PROGRAM OUTCOMES OR TO TRANSLATE THE MAIN SOCIAL COSTS AND BENEFITS INTO DOLLARS. STUDIES OF PRETRIAL RELEASE AND JUVENILE DIVERSION PROGRAMS SUGGEST THAT EVALUATION RESEARCH REMAINS THE BEST AVAILABLE METHOD FOR ASSESSING EFFECTS OF SOCIAL PROGRAMS. A DISCUSSION OF AN ANALYSIS OF FLORIDA'S CORRECTIONS PROBLEMS CONCLUDES THAT SUCH ANALYSIS HAS NOT MADE A DIFFERENCE TO THE POLICYMAKING PROCESS. TYPES OF RESEARCH FUNDED BY THE NATIONAL INSTITUTE OF LAW ENFORCEMENT AND CRIMINAL JUSTICE RESEARCH ARE EXAMINED AND THE ADVANTAGES AND DISADVANTAGES OF BASIC AND APPLIED RESEARCH ARE COMPARED. A REPORT ON RESEARCH BY THE POLICE FOUNDATION COMMENTS CRITICALLY ON THE NATURE AND RESULTS OF MUCH CRIMINAL JUSTICE RESEARCH. IT ARGUES THAT RESEARCH SHOULD ESTABLISH MORE REALISTIC PUBLIC EXPECTATIONS OF THE POLICE. FINALLY, DEDUCTIVE POLICY EVALUATION IS EXAMINED WITH EMPHASIS ON ITS APPLICATION TO LEGAL POLICY CHANGES IN THE JUDICIAL PROCESS. PREIMPACT ANALYSIS IS CONTRASTED WITH POSTIMPACT ANALYSIS. EXAMPLES INCLUDE USE OF SINGLE ANALOGY DEDUCTION TO ANALYZE THE IMPACT OF JUDICIAL PROCESS CHANGES ON PLEA BARGAINING SETTLEMENTS AND USE OF DOUBLE ANALOGY DEDUCTION TO ASSESS THE IMPACT OF JURY SIZE ON THE PROBABILITY OF CONVICTION. ALTHOUGH NOT ALL DEDUCED CONCLUSIONS CAN BE EMPIRICALLY TESTED, LOGICALLY VALID DEDUCTIONS CAN BE PRODUCED FROM EMPIRICALLY TESTED PREMISES. THIS METHOD HOLDS MUCH PROMISE FOR FUTURE EVALUATION AND IMPACT RESEARCH. REFERENCE NOTES FOR EACH PAPER ARE INCLUDED. (CFW)