We propose to evaluate the dynamical states, radiation budget, and cloud properties of the NCEP Climate Forecast System Reanalysis (CFS-R) and the ECMWF interim reanalysis (ERA-I). (As time permits, we will also evaluate the NASA GMAO MERRA.) We will begin by comparing global TOA and surface radiation budgets for both shortwave and longwave radiation from the reanalysis output with data from CERES (at TOA) and the NASA/GEWEX Surface Radiation Budget (SRB) project on monthly to interannual timescales. We will then perform a cluster classification of atmospheric states for both CFS-R and ERA-I at the DOE SGP site and the DOE TWP site in Darwin, Australia. We will then compare (a) state classification results and probability of occurrence statistics from the two reanalyses, (b) cloud occurrence profiles for each reanalysis state with each other and observed cloud profiles, and (c) probability distributions of reanalysis physical variables (such as precipitation, cloud liquid water path, and cloud radiative effect) by state with observations. We intend to link simulator codes for MISR, MODIS, and ISCCP to the reanalysis output streams to produce simulated cloud top heights (or pressures) and optical depths. Joint histograms from the reanalyses will then be compared with similar histograms from observations on a regional basis. Our final product will be a summary statement of the accuracy of the reanalysis radiation budgets, their relative similarity of dynamical states, the accuracy of cloud profiles and other geophysical variables associate with these states, and the influence of cloud errors on radiative budget errors.