New York City funds and monitors an extensive array of alternative sanctions and has been sponsoring research to develop and apply such models to improve the planning and evaluation of these programs. The models use quantitative data on cases and defendants, as well as qualitative data regarding the strength of a case. The models were applied to both drug defendants and non-drug defendants in New York City. The analysis revealed several factors with statistically significant effects on prison length. Results suggested that alternative sanctions programs should consider incorporating the initial plea offer into their screening criteria, if they do not already do so. Using the initial plea offer would differ from using the last best plea offer that New York City's alternative programs have used to estimate prison displacement. The findings also indicate that although statistical models greatly improved the accuracy of predictions of sentence length, they are imperfect and should be applied with caution. Nevertheless, any jurisdiction can replicate the models of prison sentence lengths to aid in estimating future expenditures, assessing cost effectiveness, planning future jail and prison space needs, and choosing the direction of corrections policies. Tables, notes, and 21 references (Author abstract modified)