In the decades following the education reforms in the sixties in the United States, it became obvious that the practice of evaluation had apparently not produced much useful knowledge. This state of affairs led some social scientists to declare that evaluation research was obsolescent. Others, in a more positive spirit, began the process of critical self-monitoring concentrating on the technical adequacy of evaluations. Finally, a third development took place which is indicated in the emergence of new modes of evaluation commonly characterized as “humanistic” or “naturalistic”. While these expressions of professional concern are a positive development in the debate on the adequacy of “mainstream” evaluation, they fall short in one significant respect. They do not examine explicitly the nature of quantitative evaluation theory and the assumptions upon which it is based. It is argued here that the unsatisfactory state of affairs in contemporary program evaluation cannot be resolved by addressing primarily its technical adequacy. Rather, the problems of program evaluation, as an example of educational research, are of a theoretical, and specifically epistemological, kind. In particular, it is contended that (1) evaluation, whether quantitative or qualitative, suffers from the view that there are secure foundations for our knowledge; (2) that there are no such foundations, and that consequently, evaluation must rid itself of foundationalism on pain of remaining theoretically incoherent and practically irrelevant, and (3) that the solution to these problems is to be found in a nonfoundational, materialist-pragmatist theory of knowledge which is theoretically coherent and of practical relevance for evaluation.