The offenders had been considered fit for discharge according to the criteria based on the (1959) Mental Health Act for England and Wales. The variables most highly related to outcomes were previous convictions, previous psychiatric admissions, and type of offense (homicide or property offense/nonhomicide assault). Using stepwise multiple regression, prediction instruments were developed for the prediction of subsequent assaults, court appearances, psychiatric admissions, and time in the community. Of the 125 subjects, 13 subsequently committed a violent offense. The instrument for predicting subsequent assaults was able to identify approximately half of the reoffenders. Other outcomes, subsequent court appearances, and time in the community, but not subsequent psychiatric admissions, could be predicted with more accuracy. Such instruments can be helpful in identifying those particularly prone to fail, deciding whom to discharge, developing predischarge treatment programs, and allocating aftercare resources to those most in need. Tabular data and 35 references are included.