PURPOSE: This systematic review aimed to identify risk factors for prospectively ascertained falls, focusing on those that are potentially modifiable (physical and neuropsychological factors), in older people with cognitive impairment living in the community. RESULTS: A comprehensive search of five databases identified 16 high quality (Newcastle-Ottawa Scale ≥8/9) relevant articles. Meta-analyses were undertaken for five potential fall risk factors. Of these, fallers had significantly poorer balance (standardized mean difference = 0.62, 95 %CI 0.45, 0.79) with low heterogeneity. Global cognition was not significantly associated with faller status in a meta-analysis with low heterogeneity. Meta-analyses of mobility (Timed Up-and-Go), gait speed and depressive symptoms had high heterogeneity and were not statistically significant or were borderline significant (p = 0.05). Sensitivity analyses (removing one study sample's results that differed markedly from the other included samples) reduced heterogeneity to 0% and revealed fallers had significantly poorer mobility and more depressive symptoms than non-fallers. Fallers also walked significantly slower, but heterogeneity remained high. CONCLUSIONS: In older people with cognitive impairment, fallers presented with balance deficits, poor mobility, slow gait speed and depressive symptoms. Reduced global cognition was not associated with falls. These findings suggest that interventions should target balance impairment and reveal that more high-quality research is needed.