The involvement of mental health professionals in determinations of dangerousness is both common and controversial. Among the various contexts for these evaluations, the release of potentially violent forensic patients from maximum security facilities evokes justified concern from involved experts and apprehension to outrage from the immediate community. We sought to examine how conclusions are reached on dangerousness at two sequential stages: clinical recommendations and Manifest Dangerousness Hearings decisions. In an archival study of 245 patients, we found that lack of progress in the institution and physical assaultiveness were the strongest correlates with dangerousness. In contrast, experts and review boards appeared to be relatively less influenced by diagnosis, types of treatment, and sociodemographic variables.