Most NGRI (not guilty by reason of insanity) acquitties are hospitalized for some period of time following acquittal, which raises the question of when an individual can be safely released into the community. The conditional release (CR) of persons acquitted by reason of insanity, therefore, provokes the question of public safety. This study examines the CR systems in four states--Connecticut, Maryland, New York, and Ohio. A study sample of 529 persons acquitted as NGRI from 1985 to 1987 was followed up for at least five years to determine who is conditionally released. Following a description of the CR systems, findings suggestive of the role of dangerousness and diagnosis as predictors of CR are presented. Personal characteristics are also significant factors in predicting who will be released. The length of hospitalization for this population and other descriptive factors such as history of hospitalization, arrests, substance abuse, family violence, and living arrangements are also addressed.