BACKGROUND: Sleep deprivation has been linked to higher dementia risk, but the role of weekend recovery sleep (WRS) in mitigating this risk remains unclear. This study aims to evaluate the association between WRS and dementia risk. METHODS: This prospective cohort study followed 88,592 dementia-free adults aged 40-79 years from the UK Biobank, using wrist accelerometers to measure average weekday and weekend sleep durations. Incident dementia (all-cause dementia, Alzheimer's disease [AD], vascular dementia [VaD], and nonspecific dementia) was determined using medical records. Associations were estimated using Cox proportional hazards models and restricted cubic splines (RCS). RESULTS: Among 88,592 participants (mean [SD] age, 61.9 [7.9] years), 735 (0.83%) developed dementia, including 308 (0.35%) cases of AD, 137 (0.15%) cases of VaD, and 319 (0.36%) cases of nonspecific dementia. RCS analyses revealed optimal weekday sleep durations associated with the lowest dementia risk: 8.38 h (HR, 0.73; 95% CI 0.64-0.84) for all-cause dementia, 8.33 h (HR, 0.72; 95% CI 0.58-0.89) for AD, and 9.07 h (HR, 0.59; 95% CI 0.40-0.88) for VaD. In the suboptimal sleep group (weekday sleep less than optimal duration), longer WRS was associated with reduced risks of all-cause dementia (HR, 0.801; 95% CI 0.717-0.893) and VaD (HR, 0.747; 95% CI 0.612-0.91). However, in the prolonged sleep group (weekday sleep exceeding optimal duration), longer WRS was linked to an increased nonspecific dementia risk (HR, 1.291; 95% CI 1.087-1.533). CONCLUSION: Adequate WRS can reduce dementia risk, particularly for VaD, after insufficient weekday sleep, highlighting the importance of appropriate WRS for cognitive health.