Analysts intent on developing a criminal justice system model should take into account the decisionmaker's basic needs, the problem that needs to be addressed, and the policy program changes that are to be undertaken. Simulation models include both detail models and simple models. The decisionmaker needs to know the way in which the model is used, the kinds of data in the model, and limitations that the data place on the model's use. The model's final use is based on data availability and reliability. Issues in model selection are the level and type of model use, the technical quality of the data, and the mode operation (batch or interactive). One type of simulation model is JUSSIM (linear or feedback). The JUSSIM feedback model allows more flexibility than its linear model in data analysis but requires estimates on recidivism and 'virgin' arrests (first time offenders). Using this type of systems approach in criminal justice encompasses policy planning, information-simulation, identity systems, intergovernmental liaison, and highlighting of areas that need further investigation. Canadian analysts used JUSSIM to predict inmate populations and the impact of the immigration rate on the criminal justice system and will use the model in the future to examine changes in young offenders and to study corrections jurisidiction and objectives. JUSSIM will not provide answers to problems but can provide data to be used in decisionmaking. The JUSSIM linear model is detailed, including its parameters (branching ratios, unit workloads, annual unit resource availability, resource unit costs, resource capacity constraints, and preference flow).