A multiple regression model capable of forecasting the number of persons received for supervision has 5 predictor variables and will forecast 1 year ahead without having to forecast the predictor variables. The variables are the number of persons sentenced for 13- to 35-month terms, the number of persons sentenced to probation lagged 1 year and 2 years, the number of filed criminal cases lagged 3 years, and the statistical year ended June 30 of the year forecasted. The best predictor variables for forecasting the number of persons removed from supervision next year are the numbers currently under supervision and those under supervision in the prior 1, 2, and 3 years. These variables also are useful in forecasting the number of persons under supervision. Two forecasting models for persons under supervision are outlined. A third forecast for persons under supervision can be created by subtracting the forecast of persons removed and adding the forecast of persons received to the previous year's number under supervision. Short-term forecasts can be made by using the annual total every quarter as the variable which is forecast. If there has been a straight-line trend, then a linear regression permits extrapolation of this trend. If the trend is curved, parabolic regressions can be used to extrapolate a quarter or two ahead. Considerations in forecasting are discussed. These include the importance of graphing data, the trial and error nature of the process, the need for 20 to 25 years of historical data, back lagging, and serial correlation. Problems specific to probation data and their forecasting also are discussed. A list of statistics texts and 4 references are provided.