Used step-wise multiple regression procedures to predict length of hospitalization of 225 defendants acquitted by reason of insanity in New York state. Of the 21 variables considered, only 9 (severity of offense, sex, marital status, days prior imprisonment, homicide offense, days previous civil hospitalization, educational level, race, number of victims) contributed to the significance of the regression equation. However, these accounted for but 11% of the observed variance.