In this study an inference scheme is developed to derive surface, Top of the Atmosphere (TOA), and atmospheric spectral shortwave (SW) radiative fluxes for implementation with observations from the Moderate Resolution Imaging Spectroradiometer (MODIS) type of sensors. The model takes into account all atmospheric constituents and addresses the characteristics of water and ice clouds and the variation of cloud particle effective radius. The near infrared spectrum is divided into three bands to better represent the spectral variation of cloud optical properties and water vapor absorption. A multi-layered structure allows for the treatment of surface elevation effects and for the representation of the vertical distribution of the radiative fluxes. Spectral fluxes such as Photosynthetically Active Radiation (PAR) and near-infrared radiation (NIR) are also estimated. The new inference scheme is implemented with MODIS one degree products as well as with the 5 km swath products. The derived fluxes are evaluated against the globally distributed Baseline Radiation Network (BSRN) measurements and compared with products from independent satellites. It was demonstrated that the MODIS products are in good agreement with ground observations and provide improved estimates of radiative fluxes than the other evaluated satellite products. In problematic areas for most satellite retrievals, such as the Tibet Plateau and Antarctica, the MODIS results have shown a substantial improvement. Availability of the high resolution swath based estimates of surface radiative fluxes allow, for the first time, to address unique space-time coupling issues.