Testimony by psychiatrists and other mental health professionals is used not only in civil commitment cases but also in the sentencing and dispositional stages of many criminal prosecutions. Recent studies indicate that mental health prediction testimony is not based upon any demonstrated ability to predict specific types of conduct. Further, where such testimony involves persons who do not exhibit symptoms of traditional mental illness, there is little merit in the intuitive suggestion that mental health professionals must have extraordinary abilities related to prediction. Such problems, however, do not justify a blanket prohibition of opinion testimony, and on balance the advantages of having sentencing authorities receive opinion testimony outweigh disadvantages, including cost. It is suggested that a mental health professional should be banned from expressing any predictive opinion more specific than that the subject poses a greater than average risk of engaging in future assaultive or otherwise criminal conduct. Testimony concerning a diagnosis of psychopathy when the issue is merely the subject's dangerousness should also be prohibited. Trial judges should thus limit testimony to comparative risk statements for jury trials. For cases heard by the judge alone, expert opinions should be restricted to live, formal testimony, thus eliminating views that would be acceptable in a written report but inadmissible as court testimony. In addition, defendants need to be assured of an opportunity to determine whether reports submitted to the court under the present system contain views exceeding the limitations of comparative risk statements. Finally, defendants should have the opportunity to contest the credibility of opinions expressed by professionals, even if the opinions are within suggested guidelines. Footnotes with references are provided in the article.