Phencyclidine (PCP) abuse has diminished since PCP's intrusion into American culture in the late 1970s. One of its legacies is the assumption that it provokes violent behavior in humans with predictable regularity. This assumption is so accepted that ingestion of the drug both accidentally and knowingly prior to committing a crime has been used as a defense in criminal trials. We reviewed 81 clinical reports of toxicity in humans published chiefly in North American medical journals. We searched for descriptions of violent behavior in these reports and subjected them to the following questions: (1) Was the violent behavior corroborated or only self-reported? (2) Was the presence of PCP confirmed by analysis of bodily fluids or postmortem tissue? (3) Was the presence of other drugs excluded by similar analysis of bodily fluids? We had planned to examine the reports to see whether clinicians sought evidence of previous violent behavior, but such an inquiry was rarely conducted. Of the hundreds of patients described, only three satisfied these criteria. Further, some of the papers offered evidence that reports of violence were exaggerated. These findings plus the pre-1970 prospective evaluation of thousands of patients with PCP, in which violence was never reported, led us to conclude that clinical and forensic assumptions about PCP and violence are not warranted.