Evaluation as a political arbitration entity is discussed in the case of a multilateral literacy development project in the fourth year of operation in a Third World country. An external evaluation team was invited to evaluate the project when conflict appeared between the funding agency (A) and the technical agency (B) over a project-related issue: the extension of an expired Instructional Materials Expert (IME) position. Agency A felt that IME tasks were completed, while B felt that the post extension was B's decision. The recommendations of the evaluation team were organizational and dealt with personnel issues: (1) an exit date to transfer project responsibility from A and B to government agents; (2) a change from project supervision to government advisor for a technical agent in preparation for B team exit; and (3) a four-month government funded extension of the IME position. The paper concludes that evaluation is political in its genesis, political as a process, and political in its effects. This means that evaluators must begin with collecting political data of relevance to the evaluation action, and must learn to take account of political intelligence in the process of generating, interpreting, and utilizing technical information. (PN)