This study investigates the ability of discriminant analysis to provid
e accurate predictions of hospital failure. Using data from the period
following the introduction of the Prospective Payment System, we deve
loped discriminant functions for each of two hospital ownership catego
ries: not-for-profit and proprietary. The resulting discriminant model
s contain six and seven variables, respectively. For each ownership ca
tegory, the variables represent four major aspects of financial health
(liquidity, leverage, profitability, and efficiency) plus county mark
etshare and length of stay. The proportion of closed hospitals misclas
sified as open one year before closure does not exceed 0.05 for either
ownership type. Our results show that discriminant functions based on
a small set of financial and nonfinancial variables provide the capab
ility to predict hospital failure reliably for both not-for-profit and
proprietary hospitals.