The records of 54 patients treated with interfragmentary screws were reviewed a mean of 2 years after surgery. Fifty-one fractures were of moderate severity (Ellis, 1950), 31 had a ratio of fracture length to tibial diameter of 2 or more and 17 were comminuted. An anatomical reduction, achieved initially in 44 fractures, was lost before union occurred in 14 (32 per cent), despite plaster splintage. Excellent results, with anatomical healing, were achieved in 30 patients (55 per cent) and good results with less than 5 degrees malalignment in 6 (11 per cent). There was a statistically highly significant association of satisfactory results with initial anatomical operative reduction of the fracture, a ratio of fracture length to tibial diameter of 2 or more and the absence of comminution. The orientation and number of screws, delayed operation, the degree of initial displacement of the fracture and the patient's age were not shown to affect the outcome. Delayed union occurred in 4 patients (7.4 per cent) and non-union in 3 (5.5 per cent). Deep infection occurred twice (3.7 per cent). As more satisfactory results could be expected for these fractures using other techniques, we conclude that this method cannot be recommended.