THE DISCUSSION PROCEEDS FROM THE FOLLOWING PERSPECTIVES: (1) PROGRAM EVALUATION IS A POLICY/MANAGEMENT TOOL; (2) VARIOUS LEVELS OF POLICY AND MANAGEMENT PERSONNEL HAVE NUMEROUS AND DIVERGENT EVALUATION INFORMATION NEEDS; AND (3) RARELY IS AN EVALUATION SO FATALLY FLAWED AS TO BE WITHOUT SOME RELEVANCE TO POLICY. THE REPORT IDENTIFIES POTENTIAL PROBLEMS IN THE CONDUCT OF PROGRAM EVALUATION SO THAT THEY CAN BE ANTICIPATED, ASSESSED, AND PREEMPTED. PITFALLS IN INTERPRETING DATA FOR ALTERNATIVE POLICY PURPOSES ARE EXAMINED. CONCERNS TO BE ADDRESSED BEFORE DATA COLLECTION BEGINS ARE ANALYZED TO MINIMIZE IMPEDIMENTS TO A SUCCESSFUL EVALUATION. DURING THE DATA ACQUISITION AND DATA ANALYSIS STAGES, CERTAIN INTERPRETATIONAL PROBLEMS MUST BE CONSIDERED, INCLUDING POTENTIAL DIFFICULTIES OF TRANSFERRING PROGRAMS TO NEW ENVIRONMENTS OR OF EXPANDING PROGRAMS. THE FINAL STAGE OF THE EVALUATION CYCLE IS DISCUSSED IN TERMS OF CONVERTING PROBLEMS INTO PRODUCTS. A SELECTED BIBLIOGRAPHY IS INCLUDED, AND THE APPENDIXES CONTAIN TECHNICAL DISCUSSIONS OF VARIABLES, CORRELATION, AND EXPERIMENTS. (AUTHOR ABSTRACT MODIFIED)