Introduction: Previous studies have suggested that sensory loss is linked to falls. However, most of these studies were cross-sectional designed, focused on single sensory loss, and were conducted in developed countries with mixed results. The current study aims to investigate the longitudinal relationship between hearing loss (HL), vision loss (VL) and dual sensory loss (DSL) with falls among middle-aged and older Chinese population over 7 years.Methods: The data was obtained from the China Health and Retirement Longitudinal Survey (CHARLS). In total, 7,623 Chinese older adults aged over 45 were included at baseline 2011 in this study. Self-reported falls and HL/VL/DSL were accepted. Other confounding variables included age, sex, BMI, educational level, marital status, various physical disorders and lifestyles. The impact of baseline sensory status on baseline prevalence of falls and incident falls over 7 years were assessed using logistic regression analyses. A logistic mixed model was used to assess the association between time-varying sensory loss with incident falls over 7 years after adjusted with multi-confounding factors.Results: Single and dual sensory loss groups had significantly higher prevalence of falls compared to no sensory loss (NSL) group (DSL: 22.4%, HL: 17.4%, VL: 15.7%, NSL: 12.3%). Baseline HL (OR: 1.503, 95% CI: 1.240–1.820), VL (OR: 1.330, 95% CI: 1.075–1.646) and DSL (OR: 2.061, 95% CI: 1.768–2.404) were significantly associated with prevalence of falls. For longitudinal observation over 7 years, baseline HL/DSL and persistence of all types of sensory loss were associated with incidence of falls. Time-varying HL (OR: 1.203, 95% CI: 1.070–1.354) and DSL (OR: 1.479, 95% CI: 1.343–1.629) were associated with incident falls after adjusted with multi-confounders, while VL was not.Conclusion: HL and DSL are significantly associated with both onset and increased incidence of falls over 7 year's observation in middle-aged and elderly Chinese population. Persistence or amelioration of sensory loss status could exert divergent influences on incidence of falls, which should be considered in the development of falls-prevention public health policies for aging population.