Residential substance abuse treatment (RSAT) programs for State prisoners are based on the logic of providing intense rehabilitation to offenders deemed most likely to recidivate. Based on experiences from the process and outcome evaluations of the South Carolina RSAT program, this study considered how the known barriers to successful program implementation interrelated with other evaluation process elements to frustrate outcome research design and execution. The study offered a conceptual framework for the specification of factors problematic to technocratic program evaluations in correctional settings. Barriers to corrections-based drug treatment were placed by type into four separate domains: (1) agency commitment, (2) implementation and therapeutic delivery, (3) research design, and (4) data quality. The model indicated both one-way and reciprocal relationships between the domains; a domain might either directly influence or compound the problems subsumed under other domains. The model was valuable to researchers because it assisted in identifying barriers that might arise during the evaluation of prison-based drug treatment programs. Research staff can anticipate that certain barriers are likely to occur when other barriers are experienced during a step-wise evaluation process. The barriers model also renders implications for researchers in prison settings. It is in the interest of research staff and the success of the evaluation to establish a trust-based rapport as early in the evaluation process as possible. Research conducted outside laboratory settings such as correctional facilities will undoubtedly entail day-to-day variability throughout the evaluation process. Evaluators will need to convince agency administrators and program staff that they must control certain factors within the experimental setting, particularly in the delivery of program services. Researchers should document unexpected incidents and attempt to understand how they impact findings, rather than interpret every identified barrier as a fatal flow to outcome assessment. 1 figure, appendix, 35 references