BACKGROUND: Findings on the link between the number of children and dementia risk are inconsistent, mostly studied in females, suggesting pregnancy-related changes may be a key factor in this association. METHODS: The Adult Changes in Thought Study is a cohort of adults aged ≥65 years from Kaiser Permanente Washington. The primary exposure was the number of children (0, 1, 2, 3 or ≥4), and the outcome was an incident dementia diagnosis. Cox proportional-hazards models were adjusted for demographic and early-life socioeconomic confounders. Models were then stratified by sex and by birth year <
1928 versus ≥1928. RESULTS: Among 4668 participants (average age at enrolment 74.1±SD 6.3 years
59% female), 967 (21%) had 0 children, 484 (10%) had one child, 1240 (27%) had two children, 968 (21%) had three children and 1009 (22%) had four or more children. We found no association between the number of children and dementia overall or after stratification by birth cohort. When stratified by sex and adjusting for confounders, having ≥4 children compared with two children was associated with a higher rate of dementia in males (HR=1.31, 95% CI 1.01 to 1.71). CONCLUSIONS: The number of children was not consistently associated with the risk of dementia. We observed a greater risk of dementia only among males who had ≥4 children, with the lower bound of the 95% CI marginally exceeding 1. These findings suggest that the number of children may influence the risk of dementia through other than pregnancy-related pathways.