Background Falls and fractures are a major problem. Objectives To investigate the clinical effectiveness and cost-effectiveness of alternative falls prevention interventions. Design Three-arm, pragmatic, cluster randomised controlled trial with parallel economic analysis. The unit of randomisation was the general practice. Setting Primary care. Participants People aged ≥ 70 years. Interventions All practices posted an advice leaflet to each participant. Practices randomised to active intervention arms (exercise and multifactorial falls prevention) screened participants for falls risk using a postal questionnaire. Active treatments were delivered to participants at higher risk of falling. Main outcome measures The primary outcome was fracture rate over 18 months, captured from Hospital Episode Statistics, general practice records and self-report. Secondary outcomes were falls rate, health-related quality of life, mortality, frailty and health service resource use. Economic evaluation was expressed in terms of incremental cost per quality-adjusted life-year and incremental net monetary benefit. Results Between 2011 and 2014, we randomised 63 general practices (9803 participants): 21 practices (3223 participants) to advice only, 21 practices (3279 participants) to exercise and 21 practices (3301 participants) to multifactorial falls prevention. In the active intervention arms, 5779 out of 6580 (87.8%) participants responded to the postal fall risk screener, of whom 2153 (37.3%) were classed as being at higher risk of falling and invited for treatment. The rate of intervention uptake was 65% (697 out of 1079) in the exercise arm and 71% (762 out of 1074) in the multifactorial falls prevention arm. Overall, 379 out of 9803 (3.9%) participants sustained a fracture. There was no difference in the fracture rate between the advice and exercise arms (rate ratio 1.20, 95% confidence interval 0.91 to 1.59) or between the advice and multifactorial falls prevention arms (rate ratio 1.30, 95% confidence interval 0.99 to 1.71). There was no difference in falls rate over 18 months (exercise arm: rate ratio 0.99, 95% confidence interval 0.86 to 1.14; multifactorial falls prevention arm: rate ratio 1.13, 95% confidence interval 0.98 to 1.30). A lower rate of falls was observed in the exercise arm at 8 months (rate ratio 0.78, 95% confidence interval 0.64 to 0.96), but not at other time points. There were 289 (2.9%) deaths, with no differences by treatment arm. There was no evidence of effects in prespecified subgroup comparisons, nor in nested intention-to-treat analyses that considered only those at higher risk of falling. Exercise provided the highest expected quality-adjusted life-years (1.120), followed by advice and multifactorial falls prevention, with 1.106 and 1.114 quality-adjusted life-years, respectively. NHS costs associated with exercise (£3720) were lower than the costs of advice (£3737) or of multifactorial falls prevention (£3941). Although incremental differences between treatment arms were small, exercise dominated advice, which in turn dominated multifactorial falls prevention. The incremental net monetary benefit of exercise relative to treatment valued at £30,000 per quality-adjusted life-year is modest, at £191, and for multifactorial falls prevention is £613. Exercise is the most cost-effective treatment. No serious adverse events were reported. Limitations The rate of fractures was lower than anticipated. Conclusions Screen-and-treat falls prevention strategies in primary care did not reduce fractures. Exercise resulted in a short-term reduction in falls and was cost-effective. Future work Exercise is the most promising intervention for primary care. Work is needed to ensure adequate uptake and sustained effects. Trial registration Current Controlled Trials ISRCTN71002650. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 34. See the NIHR Journals Library website for further project information.