The records of 243 patients with Thompson prostheses for displaced femoral neck fractures have been studied. One hundred and seven prostheses were inserted through an anterior approach and 136 by a posterior approach. The short-term results and complications in these otherwise comparable groups are discussed. The infection rate of 18.5 percent in the group operated upon by the posterior approach was thrice that after operations by the anterior route (6.5 per cent infected). Drained wounds had significantly less infection (6.3 per cent) compared with the undrained group (28 per cent infected). Other factors influencing the infection rate are discussed. The anterior approach offered greater stability. Of the twenty dislocations in the series, nineteen followed operations by the posterior approach. The mortality rate six weeks after operation was 6.5 per cent after the anterior approach and 20.6 per cent after the posterior approach. Statistical analysis significantly favours the anterior approach.