Using the universe of credit- and debit-card transactions in China during 2013-2015, this paper provides the first nationwide analysis of the healthcare cost of PM2.5. We leverage spatial spillovers of PM2.5 from long-range transport for exogenous variation in local pollution and employ a flexible distributed lag model to capture semiparametrically the dynamic response of pollution exposure. Our analysis shows significant impacts of PM2.5 on healthcare spending in both the short and medium terms. A 10 µg/m3 decrease in PM2.5 would reduce annual healthcare spending by more than $9.2 billion, about 1.5% of China’s annual healthcare expenditure.