This discussion traces the legislative, fiscal, and programmatic history of Project Follow Through and analyzes the many problems and controversies that have surrounded it. It is argued that problems have been largely a result of the attempt to combine research and development (R & D) and social action service programs. Lessons learned as a result of experience with Follow Through suggest that (1) large educational R & D programs should not be abandoned; (2) combining service delivery and R & D programs is very difficult; (3) ambiguities in authorizing legislation should be resolved; (4) a need exists for programmatically useful information on effective compensatory education approaches; (5) program effectiveness is only slightly related to cost, managerial properties primarily account for program effectiveness, structured programs are more effective with disadvantaged children than unstructured programs, and new R & D activities should stress complete projects and models; and (6) for the purpose of assessing program effectiveness, randomized or quasi-experimental designs are preferred, programs assessed should be clearly different, replications of all models studied should be made in sufficient number, and program implementation should be considered an essential factor. (RH)