Background. Most epidemiological studies investigating the health effects of air pollutants such as PM2.5 are susceptible to exposure measurement errors because they rely on a limited number of PM2.5 monitors in their study regions. Even for the U.S. where there is a relatively extensive ground monitoring network, these exposure errors can bias risk estimates downwards and widen the confidence intervals of the estimated effects. The spatial coverage of satellite data has a natural advantage over ground point measurements. When combined with external constraints, satellite-based remote sensing of ground-level air pollution has recently emerged as a potential major contributor to solving this problem. Aims. We aim at providing an overview of the current state of the science in using satellite aerosol remote sensing data products to derive spatially resolved PM2.5 estimates. Methods. We focus on the two primary approaches that been reported in recent literature, i.e., statistical calibration models that rely on ground measurements of PM2.5 to anchor the prediction surfaces, and scaling models that use satellite observations to scale PM2.5 simulation results based on atmospheric chemistry models. Results. We will compare the spatial and temporal resolution of most recent models, geographical range of their applications, and their prediction performances Conclusion. We will provide our thoughts on where this emerging field should go in the next few years.