In its 2015 decision in Elonis v. U.S., the United States Supreme Court overturned a conviction under a 1939federal statute making it a crime to communicate threats in interstate commerce. Defendant Anthony Douglas Elonis claimed that his Facebook rap lyrics fantasizing the murder of his estranged wife were purely therapeutic. His wife, the FBI and ultimately the jury, however, had taken his social media rants as serious threats of harm. On appeal, the Supreme Court ruled that the trial court erred in instructing the jury to apply a negligence standard (what a "reasonable person" would have perceived as a threat) to a criminal case (in which the defendant's mental state is critical). The question for the jury, ruled the Supreme Court, should have been whether Elonis actually intended to threaten or-at the very least-exhibited reckless disregard for whether his Facebook postings would be perceived as a threat. These have become questions psychologists may be called upon to consider in evaluation of individuals in forensic settings in which statements on social media are at issue. Pursuant to the 1976 decision in Tarasoff v. Regents, psychologists in California and many other states risk liability if they fail to warn potential victims of dangers posed by their clients. While the potential deadly consequences of underestimating threats may lead mental health and law enforcement professionals to err on the side of caution, over-prediction of violence has serious consequences as well-in terms of damage to therapeutic relationships, personal liberty and free speech. What does the research say about how accurately we are able to assess threat? How should the courts apply that research in constructing appropriate standards in both criminal and civil cases in an age of social media, where a lack of informative context has raised a whole new set of issues.