Causal models provide great promise for producing long- term forecasts of admissions to juvenile correctional institutions. They permit the modeling of policy, the incorporation of theoretical knowledge, and the inclusion of widely different variables and sources of information under a single theoretical framework. Three approaches to such modeling are investigated: multiple regression, a queueing network model, and a discrete-event simulation model. These approaches are compared and contrasted based on their validity (ability to predict calendar year 1992 admissions to the Ohio Department of Youth Services), ease of use, and ability to model policy. Multiple regression models select independent variables or predictors on the basis of their presumed ability to forecast the dependent variable. In this case the dependent variable of interest is the admissions of new commitments to Ohio Department of Youth Services institutions. The three categories of independent variables are demographic, socioeconomic, and policy. The queueing network model of the admissions process involves conceiving of the process as a collection of interacting queues (service systems equipped with a waiting room) known as a "queueing network." Queueing theory informs the construction and analysis of mathematical models of systems that provide service to consumers whose arrival times and service requirements are random. The discrete event simulation model is still being tested for validity. The network queueing model was found to be superior to multiple regression for predicting admissions. Simulation models are deemed to be superior in their ability to model policy. Graphs, tables, and 65 references