Objective: to determine the effect of risk factor modification and balance exercise on falls rates in residential care homes. Design: cluster randomised controlled trial. Participants: 196 residents (aged 60 years or over) in 20 residential care homes were enrolled (38% response rate). Homes were randomly allocated to intervention and control arms. A total of 102 residents were consigned to the intervention arm and 94 to the control arm. Intervention: a multifactorial falls prevention programme including 3 months gait and balance training, medication review, podiatry and optometry. Main outcome measures: number of falls/recurrent falls per person, number of medications per person, and change in Tinetti gait and balance measure. Results: in the intervention group there was a mean of 2.2 falls per resident per year compared with 4.0 in the control group; this failed to reach statistical significance (P = 0.2) once the intra-cluster correlation (ICC, 0.10) had been accounted for. Several risk factors were reduced in the intervention arm. Conclusions: falls risk factor reduction is possible in residents of care homes. A modest reduction in falls rates was demonstrated but this failed to reach statistical significance.