BACKGROUND: gait analysis is a recommended geriatric assessment for falls risk and sarcopenia; however, previous research utilises measurements at a single time point only. It is presently unclear how changes in gait over several years influence risk of recurrent falls in older adults. METHODS: we investigated 135 female volunteers (mean age ? SD: 76.7 ? 5.0 years; range: 70-92 years) at high risk of fracture. Gait parameters (speed, cadence, step length, step width, swing time and double support phase) were assessed using the GAITRite Electronic Walkway System at four annual clinics over 3.7 ? 0.5 years. Participants reported incident falls monthly for 3.7 ? 1.2 years. RESULTS: increasing gait speed (odds ratio: 0.96; 95 confidence interval 0.93, 0.99) and step length (0.87; 0.77, 0.98) from baseline to final follow-up was associated with reduced likelihood of being a recurrent faller over the study period. No significant associations were observed for baseline gait parameters (all P = 0.05). At the second follow-up (2.8 ? 0.6 years), an increase in swing time (0.65; 0.43, 0.98) was associated with reduced likelihood, while an increase in double support phase (1.31; 1.04, 1.66) was associated with increased likelihood, for being a recurrent faller in the subsequent 1.3 years following this time point. CONCLUSION: changes in gait parameters over several years are significantly associated with the likelihood of being a recurrent faller among community-dwelling older women at high risk of fracture. Further research is required to develop gait monitoring guidelines and gait parameter decline cut points that may be utilised by clinicians to identify older adults at risk of incident falls and sarcopenia.