AbstractIntroductionThere is inconsistent evidence on the associations of sleep duration and daytime napping with dementia risk.MethodsIn the Million Women Study, a total of 830,716 women (mean age, 60 years) were asked about sleep duration (<7, 7–8, >8 hours) and daytime napping (rarely/never, sometimes, usually) in median year 2001, and were followed for the first hospital record with any mention of dementia. Cox regression estimated dementia detection risk ratios (RRs) during 17‐year follow‐up in 5‐year intervals.ResultsWith 34,576 dementia cases, there was strong attenuation over follow‐up in the RRs related to long sleep duration (>8 vs 7–8 hours) and usually napping (vs rarely/never). Short sleep duration was modestly, positively associated with dementia in the long term (RR = 1.08, 95% confidence interval [CI] 1.04–1.12).DiscussionThere was little evidence to suggest that long sleep duration and regular napping are associated with long‐term dementia risk. Short sleep duration was modestly associated with dementia risk, but residual confounding cannot be excluded.Highlights Long sleep duration was not associated with long‐term dementia risk. Daytime napping was not associated with long‐term dementia risk. There is some evidence for a small higher risk of dementia related to short sleep.