Two studies (n = 43,992 participants) were included. In Zhou et al. (2020), 793 strokes occurred among 19,742 nappers versus 325 strokes among 12,008 non-nappers. In Wu et al. (2024), 1,177 strokes occurred among 8,702 nappers versus 317 strokes among 2,550 non-nappers. The fixed-effect model indicated a significant association between napping and stroke (OR = 0.258; 95% CI: 0.165–0.351; p < 0.001). However, the random-effects model showed a non-significant association (OR = 0.253; 95% CI: –0.052 to 0.558; p = 0.10), with low heterogeneity (I² = 0.9%). Fail-safe N analyses suggested limited robustness (Rosenthal’s N = 0; Orwin’s N = 2), indicating susceptibility to publication bias.