This paper concerns a national evaluation study of the Upward Bound (UB) Program conducted for the U.S. Office of Education by the Research Triangle Institute. It describes the practical constraints imposed on the study design and the features of the design that were shaped by these constraints. The design features include the quasi-experimental and cross-sectional approaches. Also considered are methodological problems presented by the design and measures taken to alleviate them, including a synthetic cohort approach and the choice of the comparison group. (Author/RC)