Use of seclusion and restraint in 23 adult public psychiatric hospitals in New York State was examined by comparing demographic and diagnostic characteristics of 657 patients who were secluded or restrained during a four-week period with characteristics of 22,939 patients who were not. Logit analysis was used to calculate the probability of seclusion and restraint of individual patients with various combinations of characteristics. Characteristics associated with high probabilities included age less than 26 years, length of stay from 30 to 365 days, involuntary legal status, female gender, a diagnosis of mental retardation, and residence in a hospital with a high rate of seclusion and restraint. The rate of seclusion and restraint in the 23 hospitals ranged from .4 to 9.4 percent of patients. Both patient characteristics and the hospital of residence were needed to explain the case-by-case probability of a patient's being secluded or restrained.