One hundred and forty-three patients presenting with ankle sprains within 24 h of injury were entered into a double blind study. Treatment consisted of a standardized regime of high dose non-steroidal anti-inflammatory medication and an elastic support for all patients, who were then randomly allocated to two groups. One group received immediate cold therapy, the other received simulated therapy. Assessments made at 7 days showed a trend in favour of the group receiving cold therapy, although this did not reach significance. It is concluded that cold therapy together with compression may have a beneficial effect but that a single application in the accident and emergency department is not justified when a background therapy of non-steroidal anti-inflammatory medication is given.