The MCMI-III was useful in highlighting the level of psychological difficulty as reported by the residents; the PCL-R scores differed between offense groups, but proved to be the most useful tool in predicting overall failure. For the MCMI-III, results should be interpreted with caution as it is not an objective measure of PD. Nevertheless, the resident group reported levels of antisocial traits and drug dependency, which were consistent with independent reports and records, which supported its validity. Some of the scores indicated a likely need for clinical intervention; for example, the men reported experiencing high levels of negative affect, and endorsed depressive and borderline traits to a significant level. In most respects, the MCMI-III was neither a strong predictor of behavior within the hostel nor of overall failure; these indicators of depressive features might suggest that the association between self-defeating personality traits and worse outcome might be clinically robust and of practical importance. In terms of the PCL-R, low scores indicated a high probability of success since two-thirds of moderate scorers succeeded. About one-third of high scorers succeeded, giving the impression that there were two types of psychopaths: those who presented as stable within the hostel but might have posed a longer-term risk in the community, and those who were unstable within the hostel and who were characterized by high levels of negative affect. Data were collected from 93 high-risk residents (with criminal convictions) who were admitted between April 1999 and February 2001 to a specialist hostel which is managed by the probation service and includes a high level of forensic mental health service input. Tables and references