THIS ARTICLE IS THE RESULT OF A SYMPOSIUM WHICH INCLUDED REPRESENTATIVES FROM THE DEPARTMENT OF COMMERCE, THE DEPARTMENT OF LABOR, THE ENVIRONMENTAL PROTECTION AGENCY, THE LAW ENFORCEMENT ASSISTANCE ADMINISTRATION (LEAA), THE NATIONAL INSTITUTE OF MENTAL HEALTH, THE OFFICE OF EDUCATION, AND THE OFFICE OF HEALTH, PLANNING, AND EVALUATION. IT EXAMINES THE EXPERIENCE OF EACH OF THESE FEDERAL AGENCIES WITH EVALUATION AND PROBLEMS THEY HAVE ENCOUNTERED. THE INITIAL OVERVIEW IDENTIFIES THREE PERSPECTIVES ON EVALUATION--MANAGEMENT, KNOWLEDGE, AND ACCOUNTABILITY. EACH IS EXAMINED IN DETAIL AND THE CONFLICTS WHICH ARISE WHEN CONGRESS MANDATES AN ACCOUNTABILITY PERSPECTIVE WHILE THE AGENCY WANT A MANAGEMENT PERSPECTIVE ARE DETAILED. SIX USERS OF DATA WERE IDENTIFIED: FEDERAL AGENCY POLICY MAKERS AND PROGRAM MANAGERS; THE OFFICE OF MANAGEMENT AND BUDGET; CONGRESS AND THE GENERAL ACCOUNTING OFFICE; RESEARCHERS, PLANNERS, AND EVALUATORS; STATE AND LOCAL POLICY MAKERS; AND THE PRESS AND THE PUBLIC (ESPECIALLY IN THE CASE OF HIGHLY VISIBLE AND POLITICALLY IMPORTANT ASSESSMENTS). THE DATA NEEDS OF EACH GROUP ARE EXAMINED. AN AGENCY-BY-AGENCY REVIEW OF EVALUATION POLICY AND EXPERIENCE IS THEN GIVEN. LEAA'S FOCUS ON ATTEMPTS TO FIND APPROPRIATE MEASURES FOR ATTEMPTS TO SOLVE COMPLEX PROBLEMS ARE EXAMINED IN DETAIL. THE CONCLUSION EMPHASIZES THE IMPORTANCE OF THE PROGRAM ADMINISTRATOR IN ANY EVALUATION. REFERENCES ARE PROVIDED. (GLR)