Objective: to investigate how physiological and cognitive factors mediate the association between CSRT performance and multiple falls by use of path analysis. Methods: 294 retirement-village residents, aged 62-95 years, undertook CSRT tests, requiring them to step onto one of four randomly illuminated panels, in addition to physiological and cognitive tests. Number of falls was collected during 1-year follow-up. Results: 79 participants (27%) reported two or more falls during the follow-up period. Regression analyses indicated CSRT was able to predict multiple falls by a factor of 1.76 for each SD change. The path analysis model revealed that the association between CSRT and multiple falls was mediated entirely by the physiological parameters reaction time and balance (postural sway) performance. These two parameters were in turn mediated over a physiological path (by quadriceps strength and visual contrast sensitivity) and a cognitive path (cognitive processing). Conclusions: this study provides an example of how path analysis can reveal mediators for the association between a functional measure and falls. Our model identified inter-relationships (with relative weights) between physiological and cognitive factors, CSRT and multiple falls.