Over the past decade, the ability of the mental health professional to assess the likelihood of dangerous behavior in the mentally disordered offender has been subjected to severe criticism. The accuracy of clinical judgment in this regard appears doubtful in terms of several investigations which show a large number of false positive predictions. Most of the empirical studies in this area, to date, exhibit serious methodological flaws (e.g., lack of appropriate contrast groups, use of inferred rather than stated clinical predictions, variable length of the followup period, failure to examine the interclinician reliability of predictions) which cast doubt upon the claim that mental health professionals often fail in their attempts to predict dangerous behavior. Recommendations are offered for use in the design of a methodologically sound study which would permit stronger conclusions to be drawn with respect to the clinician's ability or inability to make predictions. Footnotes, 17 references, and an appended summary chart are provided. (Author abstract modified)