OBJECTIVES: To compare internal fixation, bipolar hemiarthroplasty and total hip arthroplasty for the management of displaced subcapital fracture of the hip in previously fit patients of 60 years or older. DESIGN: A prospective randomised clinical trial. SETTING: This multicentre trial was carried out in 11 Scottish hospitals with acute orthopaedic trauma units. PARTICIPANTS: The participants were 298 previously fit patients of 60 years or older with displaced subcapital hip fractures. INTERVENTIONS: The three surgical interventions for comparison were reduction and fixation, bipolar hemiarthroplasty and total arthroplasty (total hip replacement). Participating surgeons elected to randomise patients either among all three types of operation (three-way randomisation) or just between fixation and hemiarthroplasty (two-way randomisation). MAIN OUTCOME MEASURES: Clinical outcomes were mortality rates, reoperation rates and the complication rates associated with each procedure. Functional outcome was measured using a hip specific questionnaire [Johanson Hip Rating Questionnaire (HRQ)] and a general health status questionnaire [EuroQol 5 Dimensions (EQ-5D)]. Economic analysis compared the costs in the randomised groups of hospital treatment for the initial and subsequent admissions for up to 2 years. RESULTS: Altogether, 207 patients were randomised among all three trial operations, and 91 between just fixation and bipolar hemiarthroplasty. There were no statistically significant differences in clinical outcomes, but confidence intervals (CIs) were wide. At 2 years fixation failure reached 37% among those allocated fixation and 39% had undergone further surgery. Further surgery rates after hemiarthroplasty and total hip replacement were 5% and 9%, respectively. The group allocated fixation had significantly worse HRQ and EQ-5D scores than both arthroplasty groups at 4 and 12 months. At 24 months the results still favoured arthroplasty, but the overall HRQ and EQ-5D scores were no longer statistically significant. Total hip replacement had the best patient-assessed outcome scores. At 24 months the overall HRQ and EQ-5D scores for total hip replacement were significantly better than for hemiarthroplasty. The mean costs for the initial episode ranged from 6384 pounds Sterling for fixation to 7633 pounds Sterling for total hip replacement. The cost differences were largely due to differences in theatre costs and the cost of prostheses and hardware. The cumulative cost over 2 years of hemiarthroplasty was around 3000 pounds Sterling lower than for fixation (95% CI 1227 pounds Sterling to 7192 pounds Sterling). Compared with total hip replacement, both fixation and hemiarthroplasty were characterised by increased costs arising from hip-replacement admissions. When total (initial episode and subsequent hip-related admissions) hip-related costs are compared, total hip replacement conferred a cost advantage of around 3000 pounds Sterling per patient (versus hemiarthroplasty, 95% CI -pounds Sterling 1400 to 7420 pounds Sterling). CONCLUSIONS: In fit, older patients the results of the study show a clear advantage for arthroplasty over fixation; arthroplasty was more clinically effective and probably less costly over a 2-year period postsurgery. The results suggest that total hip replacement has long-term advantages over bipolar hemiarthroplasty, but these findings are less definite. This study provided support for the use of total hip replacement to treat displaced intracapsular hip fractures in fit, older patients. A larger trial comparing total versus hemiarthroplasty for these fractures could help to verify these findings. It would also be useful to know whether the findings of this study apply to patients aged 60 years or less who are usually treated with reduction and fixation. A clinical trial comparing arthroplasty versus fixation in patients older than 40 years would be a logical extension of the current study.