In the contentious debate over the use of randomized clinical trials (RCTs) in education research, little attention has been given to how methodological issues arise and/or are resolved in the implementation of random assignment. Following a review of the methodological-theoretical literature on RCTs and alternative research designs, this article analyzes qualitative and quantitative data on three years of random assignment in a multi-site literacy intervention for adolescents. Practical lessons are drawn about how researcher practices and learning by school staff contributed to random assignment compliance. Consideration of challenges encountered in implementing random assignment suggests that 1) researcher communication with program staff improves compliance, but may not overcome the need for learning through experience; 2) in keeping with arguments in favor of random assignment-based research, random assignment may control for diverse selection processes that would undermine valid treatment-control comparisons in some non-RCT research designs; and 3) in keeping with arguments against random assignment-based research, random assignment may lead to samples dissimilar to target populations. In concluding, the broader limitations of research that focus on a priori causes and quantifiable effects are noted. (Contains 2 tables and 1 note.)