Introduction and aim There are different methods of internal fixation of intracapsular fractures of the hip of which three AO screws is one of the more popular. There have been no evidence-based publications describing the optimal position for screws. The aim was to establish the relationship between screw position and angle, and subsequent failure of union.

Method Using computer software we studied the position of AO screws in 395 consecutive patients inserted between 1989 and 2003. Follow-up was prospective and for a minimum of 100 days. The diagnosis of non-union was made clinically and confirmed radiographically.

Results The mean age of our population was 73.9 years (range 22–96). Eighty-six (21.8%) were male. Three hundred and twenty seven (82.8%) came from their own home. The mean time between fall and surgery was 37.0 hours and between admission and surgery 20.9 hours. The mean length of radiographic follow-up for those fractures that did not develop non-union was 454 days (range 94–1898). Of the 395 patients 242 (61%) fractures united and 153 (39%) fractures suffered non-union.

Radiographic analysis suggests that the position of the screws on the AP view (superior, middle, inferior or spread) did not alter the outcome significantly. However three factors were related to lower risk of non-union on the lateral view: the closer the middle screw to the center of the head (p< 0.04), the more anterior the anterior screw (p< 0.008), and the greater the ‘spread’ between the anterior and posterior screws (p< 0.005).

Conclusions We conclude that to reduce the risk of non-union with screw fixation of intracapsular fractures of the hip, in the lateral view the middle screw must be positioned as close to the centre of the femoral head as possible, and the anterior and posterior screws achieve maximal spread.