While exposure to nicotine during developmental periods can significantly affect brain development, studies examining the association between maternal smoking and autism spectrum disorder (ASD) in offspring have produced conflicting findings, and prior meta-analyses have found no significant association. Our meta-analysis used a novel approach of investigating population-level smoking metrics as moderators. The main meta-analysis, with 22 observational studies comprising 795,632 cases and 1,829,256 control participants, used a random-effects model to find no significant association between maternal smoking during pregnancy and ASD in offspring (pooled odds ratio (OR) = 1.16, 95% CI: 0.97-1.40). However, meta-regression analyses with moderators were significant when we matched pooled ORs with adult male smoking prevalence (z = 2.55, p = 0.01) in each country, using World Health Organization data. Our study shows that using population-level smoking metrics uncovers significant relationships between maternal smoking and ASD risk. Correlational analyses show that male smoking prevalence approximates secondhand smoke exposure. While we cannot exclude the possibility that our findings reflect the role of paternal or postnatal nicotine exposure, as opposed to maternal or in utero nicotine exposure, this study underlines the importance of investigating paternal and secondhand smoking in addition to maternal smoking in ASD.