Dr. Volavka correctly describes in what ways studying violent behavior in humans is quite different from studying aggression in animals. Human violence is relatively rare and unpredictable, it varies widely, it usually seems related to environmental events difficult for the researcher to control, and, I would add, we don't know if violence is a unitary factor across diagnoses. All this makes drug studies difficult. Dr. Volavka brings judicious conclusions after presenting the data fairly. For example, regarding the alleged paradoxical reactions to benzodiazepines, Dr. Volavka quotes the anecdotal reports but notices that these patients had had previous violent episodes and that most reports refer to long-term treatment, whereas to treat violent behavior, most clinicians use benzodiazepines short-term. The only systematic data for paradoxical aggression from using benzodiazepines come from paper-and-pencil tests of hostility, which found that the benzodiazepine group showed more aggression than the placebo group, or that one benzodiazepine group showed more aggression than another. I concur with Dr. Volavka's conclusions that we should generalize from these tests to violent behavior only with much caution. If there are paradoxical reactions, they are rare and the effect on clinical practice is limited.