Abstract Background Longitudinal evidence of poor visual acuity associating with higher risk of incident dementia is mixed. This study aimed to examine if poor visual acuity was associated with higher dementia incidence in a large community cohort of older adults, independent of the possible biases relating to misclassification error, reverse causality, and confounding effects due to health problems and behaviors. Methods A total of 15,576 community-living older adults without dementia at baseline were followed for 6 years to the outcome of incident dementia, which was diagnosed according to the ICD-10 or a Clinical Dementia Rating of 1 to 3. Visual acuity was assessed using the Snellen’s chart at baseline and follow-up. Important variables including demographics (age, sex, education, and socioeconomic status), physical and psychiatric comorbidities (cardiovascular risks, ophthalmological conditions, hearing impairment, poor mobility, and depression), and lifestyle behaviors (smoking, diet, physical, intellectual, and social activities) were also assessed. Results Over 68,904 person-years of follow-up, 1,349 participants developed dementia. Poorer visual acuity at baseline was associated with higher dementia incidence in 6 years, even after adjusting for demographics, health problems, and lifestyle behaviors, and excluding those who developed dementia within 3 years after baseline. Compared with normal vision, the hazard ratio of dementia was 1.19 (p = .31), 2.09 (p < .001), and 8.66 (p < .001) for mild, moderate, and severe visual impairment, respectively. Conclusions Moderate-to-severe visual impairment could be a potential predictor and possibly a risk factor for dementia. From a clinical perspective, older adults with poor visual acuity might warrant further risk assessment for dementia.