Evaluations of multisite programs present special challenges to evaluators concerned with the use of evaluation results. This paper offers an approach that holds promise for mitigating the problems inherent in evaluating programs at multiple sites. A brief analysis is presented of the challenges these programs present and the reasons these challenges exist. In sum, these challenges exist because each site implements a program differently. Two federal education programs, the Educational Partnerships Program and the Star Schools distance learning program, are described and are used as examples throughout the paper. A second section describes a general theoretical approach to evaluation that is based on the theory-driven evaluation approach of H.-T. Chen (1989) and the work of Jacques Derrida (1981) and others who emphasize the deconstruction of concepts and activities. The evaluation framework, illustrated in figure 1, incorporates the elements of: (1) context; (2) problem definition; (3) program; (4) funded projects; (5) activities; and (6) outcomes. Addressing program developers' values and assumptions is particularly useful in multisite program evaluations. (Contains 42 references.) (SLD)