The project research staff requested directors of corrections to supply information on the projection techniques used by their departments; replies were received from 45 States and 5 other jurisdictions. Information requests focused on four issues, including methods used, projections for prison population in the near and distant future, accuracy of past projections, and use of projections. Primary prediction methods revealed by the survey are the linear regression method; multiple regression; ratio; simulation; and time-series, nonlinear models. Findings indicate that much of the current methodology is borrowed from other fields, particularly from economic and demographic forecasting. It is suggested that improved lines of communication among forecasters, policymakers, and other interested parties are necessary for improving methodologies in the future. In developing such methodologies, important considerations include defining the goals of the projection model, improving the data bases, and incorporating informal techniques into the more formal models. Any new method will require testing of assumptions and experimentation with the data before it can be used reliably. Extensive tables and figures, three appendixes of survey results, and about 40 references are provided. (Author summary modified)