Collocated Interagency Monitoring of Protected Visual Environments (IMPROVE) particulate matter (PM) less than 2.5 microm in aerodynamic diameter (PM2.5) chemically speciated data, mass of PM less than 10 microm in aerodynamic diameter (PM10), and Aerosol Robotic Network (AERONET) aerosol optical depth (AOD) and size distribution at Bondville, IL, were compared with satellite-derived AOD. This was done to evaluate the quality of the Geostationary Operational Environmental Satellite (GOES) and Moderate Resolution Imaging Spectroradiometer (MODIS) AOD data and their potential to predict surface PM2.5 concentrations. MODIS AOD correlated better to AERONET AOD (r = 0.835) than did GOES AOD (r = 0.523). MODIS and GOES AOD compared better to AERONET AOD when the particle size distribution was dominated by fine mode. For all three AOD methods, correlation between AOD and PM2.5 concentration was highest in autumn and lowest in winter. The AERONET AOD-PM2.5 relationship was strongest with moderate relative humidity (RH). At low RH, AOD attributable to coarse mass degrades the relationship; at high RH, added AOD from water growth appears to mask the relationship. For locations such as many in the central and western United States with substantial coarse mass, coarse mass contributions to AOD may make predictions of PM2.5 from AOD data problematic. Seasonal and diurnal variations in particle size distributions, RH, and seasonal changes in boundary layer height need to be accounted for to use satellite AOD to predict surface PM2.5.