The ability of radiative transfer models to predict broadband fluxes under clear sky conditions has been the focus of numerous studies over the past few years. Because models generally can simulate accurately the direct normal component, the emphasis is on how well the models can predict diffuse irradiance. Previous studies have found that at a high altitude site (the Mauna Loa observatory) and at sea level Arctic sites (the ARM sites at Barrow and Atqasuk Alaska) the modeled diffuse irradiance is in good agreement with the measured values. This agreement is true even in cases of moderate aerosol loading. What is puzzling, however, is that in similar studies performed at the mid-latitude ARM Southern Great Plains (SGP) site in Oklahoma the model calculations are consistently higher than the measured diffuse irradiance (where corrections for IR loss have been applied). This result has led to much speculation as to whether this observed discrepancy might be attributed to the mid-latitude location and the influence of continental air-masses or more specifically to the SGP site itself. To address this question, we are carrying out comparison studies of modeled and measured clear sky diffuse irradiance for several mid-latitude sites that are part of the Surface Radiation Budget Observing (SURFRAD) network. Here we present direct comparisons of the modeled and measured diffuse irradiance for selected clear-sky cases at four SURFRAD sites. For these cases all model inputs are obtained from sounding and instrument measurements at the respective sites. Additionally, for each site the cases are selected to show both low and moderate aerosol loading. These results demonstrate the geographical extent of the observed diffuse irradiance discrepancy.