A consecutive series of patients with all types of cervical hip fracture (both undisplaced and displaced) were randomised to osteosynthesis with Hansson hook-pins (n = 98) or AO-screws (n = 101). Background parameters, fracture type and reduction of the fracture did not differ significantly between the groups. Fifty-seven percent of the patients were operated on within 6 h of admission to hospital, 74% within 12 h and 92% within 24 h. The mean (median) time for operation was 36 (30) min for the hook-pins and 40 (35) min for the AO-screws. The devices were significantly better positioned in the hook-pin group (81% of cases good) compared to the AO-screws (66% good) (p = 0.04). In all, 72% of the patients had no deficiency either in reduction of the fracture, positioning of the implants or had drill penetration of the femoral head. Direct unrestricted weight bearing was encouraged in 92% of the hook-pin and 90% of the AO-screws group. The mean (median) hospital time was 13 (10) days with no significant difference between the groups. Following treatment, 5% walked without aids, 76% of the patients walked with some aids, and 16% could not walk. The walking ability was not known for 4%. At four months, 59% of the patients were living in their own home (64% before fracture), 18% (25% before) in a nursing home, 5% (11% before) in other accommodation and 18% were dead. After two years, 77% of the hook-pin patients had not needed any re-operation compared to 73% in the AO-screw group. In total a secondary hemi-arthroplasty had been performed in 7% and total hip arthroplasty in 12% of the patients. Extraction only of osteosynthesis material had been performed in 5%. The difference in the reoperation rates between the two methods was not significant. In the undisplaced fractures, 84% of the patients had not: needed any reoperation after two years compared to 70% among the displaced fractures. Major reoperation had been performed in 10% (1% hemi and 9% total hip arthroplasty) in the patients with undisplaced fractures compared to 26% in those with displaced fractures (10% hemi, 16% total hip arthroplasty and 1% Girdlestone operation). The remaining patients had only undergone removal of metalwork. Osteosynthesis thus proved to be a successful operation in many of the patients with displaced fractures. A preoperative, prognostic-based selection between osteosynthesis and arthroplasty is the future goal for optimised femoral neck fracture treatment.