Findings from this large Japanese cohort supported a significant inverse association between chocolate consumption and risk of developing stroke in women. However, residual confounding could not be excluded as an alternative explanation for our findings.