The politics of program evaluation are discussed from the personal perspective of the Director of the General Accounting Office's Program Evaluation and Methodology Division, which has produced reports for committees of the United States Congress. It is concluded that successful evaluations must be useful to others and must understand the political system in which evaluation operates. The political system encompasses not only executive policymakers, but also the legislative (and, perhaps, in the future, the judicial) branch, and their interactions. Evaluation planners should understand that the policy question must be of interest to the user and that the findings must answer that question. Evaluations must be credible, timely in accordance with the policymaking cycle, and cost effective. A good relationship with the executive agency implementing the evaluation's recommendations is significant. It is also suggested that evaluative thinking be introduced into the policy forum before legislation is proposed. Panels of experts strengthen an evaluation's credibility. It is important that the evaluator master a wide variety of approaches and that the user understand the evaluation design process. Editing of the evalution report, prioritization of findings, and meta-analysis or evaluation synthesis are useful. Impact questions are generally asked by legislators, but descriptive questions are not typically raised by policymakers. Taking political processes into account affects time allocation; more time must be devoted to negotiation, discussion, briefing, accuracy-checking, prioritization, and presentation. Traditional evaluation has improved as a result of the major modifications described in this presentation. (GDC)