The first section uses examples from Victoria to discuss the statistical relationships between the demographic structure of the general population and the composition of the 'clientele' of the justice and penal systems. It presents evidence to support the contention that these relationships are sufficiently stable over time to be used in projecting future client numbers. After reviewing the types of models which traditionally have been used in this kind of forecasting, a subsequent section identifies the stages of the Oracle Model. The latter model takes receivals from the court system during a year, adds them to the various correction populations on hand at the beginning of the year, and works out which of those individuals will still be under correctional treatment at the end of the year. A figure shows a flowchart of the model. Reporting model use and model results, the final section identifies the need to obtain a 'base run' and considers initial variations on the base run. It also presents an optimistic, most likely, and a pessimistic scenario and discusses implications for the security classification system of the selected scenarios and the effects of the various scenarios on noncustodial client numbers. Nine tables, 23 figures, and 23 references are provided. Appendixes contain the Fortran listing and definitions of the variables used.