The study took a cohort of hospital orders from another investigation (Gibbens, Soothill, and Pope, 1977) which collected the basic details of men and women medically remanded in magistrates' courts (for 1969) and in higher courts (for 1969 and 1970) in Inner London and in the Wessex Regional Hospital Board area. In addition, a more detailed study was made of all offenders remanded for a medical report in Wessex between September 1, 1970, and April 30, 1971. There were 247 males and 45 females in these samples who received hospital orders. The current study deals only with the 200 males whose files were traced by the Criminal Records Office. The period of followup was 3 years from the committal date; additional data were obtained for 5 years for all cases and for up to 9 years for a smaller number. The only offenses considered were those which 'caused real harm to people.' In the 3-year followup after admission, a 'serious failure' rate of 5 percent was found, compared with 2 percent in a prior similar study (Walker and McCabe, 1973). The difference is understandable in view of the fact that the sample in this study was discharged from the hospital more quickly than the other study's sample, and there was a higher proportion of 'bad risks.' The cumulative total of 'serious failures' at the end of 5 years was 10 percent, and by the end of the seventh year, 'serious failures' had risen to 15 percent. The importance of longitudinal studies in this area of research is indicated, in order to measure the potential harm to citizens occasioned by the current system of treatment and release of those under compulsory hospital orders. Tabular data, footnotes, and 15 references are provided.