Consecutive patients with a confirmed rupture of at least one of the lateral ligaments of the ankle were randomly assigned to receive either operative or functional treatment. They were evaluated at a median of 8 years (6 to 11). In total, 370 patients were included. Follow-up was available for 317 (86%). Fewer patients allocated to operative treatment reported residual pain compared with those who had been allocated to functional treatment (16% versus 25%, RR 0.64, CI 041 to 1.0). Fewer surgically-treated patients reported symptoms of giving way (20% versus 32%, RR 0.62, CI 0.42 to 0.92) and recurrent sprains (22% versus 34%, RR 0.66, CI 0.45 to 0.94). The anterior drawer test was less frequently positive in surgically-treated patients (30% versus 54%, RR 0.54, CI 0.41 to 0.72). The median Povacz score was significantly higher in the operative group (26 versus 22, p < 0.001). Compared with functional treatment, operative treatment gives a better long-term outcome in terms of residual pain, recurrent sprains and stability.