When describing the risks of surgery to a patient, there is a common and mistaken supposition by surgeons that there exists a numeric threshold of improbability beyond which there is no need to disclose. Where should the line be drawn?