Abstract Altered 24-hour rest–activity rhythms may be associated with cognitive impairment in older adults, but evidence from prospective studies is limited. Nonparametric methods were used to assess actigraphy-based activity patterns in 2 496 older men. Incident cognitive impairment was assessed 4 times over 12 years using the Modified Mini-Mental State Examination (3MS) and Trails B tests, self-reported medication use, and clinical diagnosis. The highest quartile (vs the lowest) of intradaily variability and the lowest quartiles (vs the highest) of interdaily stability and relative amplitude were associated with incident cognitive impairment (hazard ratio [95% confidence interval]: 1.82 [1.31–2.53], 1.36 [0.99–1.86], and 1.85 [1.33–2.56], respectively). A larger increase in intradaily variability over 7.5 years was associated with a greater subsequent decline in 3MS scores but not in Trails B performance. In conclusion, less stable and more variable rest–activity rhythms may represent early biomarkers of cognitive impairment in older men.