The coronavirus disease 2019 (COVID-19) pandemic is still rapidly spreading globally. To probe high-risk cities and the impacts of air pollution on public health, this study explores the relationship between the long-term average concentration of air pollution and the city-level case fatality rate (CFR) of COVID-19 globally. Then, geographically weighted regression (GWR) is applied to examine the spatial variability of the relationships. Six air pollution factors, including nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone (O-3), PM2.5 (particles with diameter <= 2.5 mu m), PM10 (particles with diameter <= 10 mu m), and air quality index (AQI), are positively associated with the city-level COVID-19 CFR. Our results indicate that a 1-unit increase in NO2 (part per billion, PPB), SO2 (PPB), O-3 (PPB), PM2.5 (microgram per cubic meter, mu g/m(3)), PM10 (mu g/m(3)), AQI (score), is related to a 1.450%, 1.005%, 0.992%, 0.860%, 0.568%, and 0.776% increase in the city-level COVID-19 CFR, respectively. Additionally, the effects of NO2, O-3, PM2.5, AQI, and probability of living with poor AQI on COVID-19 spatially vary in view of the estimation of the GWR. In other words, the adverse impacts of air pollution on health are different among the cities. In summary, long-term exposure to air pollution is negatively related to the COVID-19 health outcome, and the relationship is spatially non-stationary. Our research sheds light on the impacts of slashing air pollution on public health in the COVID-19 pandemic to help governments formulate air pollution policies in light of the local situations.