The association between maternal exposure to PM2.5 and birthweight varies geographically, which may be caused by susceptibility. Whether this population-level association is a function of mixtures of individuals with different susceptibilities is unclear. We investigated the probability distribution of individuals with different susceptibilities to PM2.5-related birthweight change, and evaluated spatial variation of the effect across the United States (US). We estimated the individual-level susceptibility using the effect of PM2.5 among a homogenous subpopulation, which was defined by a specific combination of modifiers. According to frequencies for all combinations, we derived the probability distribution of differential susceptibilities across the US and by states. From birth certificates across the US (1999–2004), we analyzed a total of 18,317,707 samples of singletons. Of the samples, 54–55% were assigned valid exposures, and linked to PM2.5. The subpopulation-specific associations of PM2.5 on birthweight change (i.e., susceptibilities) ranged from negative to positive. For the first-trimester exposure, 61.4% of the associations were negative, and the mean was −1.01 g (95% confidence interval, CI: −1.63, −0.38) of birthweight change per 5 μg/m3 increase of PM2.5. The state-level associations varied (from −2.04 g [−2.76, −1.31] in New Hampshire to −0.30 g [−1.01, 0.41] in Texas) with demographic compositions in the US. The between-state variations of maternal race and education level were the greatest contributors to the spatial heterogeneity. Our findings may be useful to the policymaker in planning interventions for subpopulations susceptible to ambient pollution.