This study examined a state-operated mental retardation service delivery system and conducted a systems analysis of the case coordinator referral system to determine the characteristics of individuals who were likely to be referred to supported employment. It also performed an outcomes analysis of the system's supported employment program to determine who was likely to be successful in the program. This examination was conducted by building prediction models using existing case file information. This study provided some evidence that empirically derived prediction models (i.e., logistic and multiple regression) can be a functional evaluation tool for supported employment managers, administrators, and rehabilitation counselors.