Cloud and precipitation simulated using the 3-dimensional Goddard Cumulus Ensemble (GCE) model are compared to TRMM TMI and PR rainfall measurements and CERES single scanner footprint (SSF) radiation and cloud retrievals. Both the model simulations and retrieved parameters are based upon observations made during the SCSMEX field campaign. The model simulated cloud and rain systems are carefully evaluated by systematically examining important parameters such as surface rain rate, convective/stratiform percentage, rain profiles, cloud properties and precipitation efficiency. It is demonstrated that the GCE model is capable of simulating major convective system development and reproduces the total surface rainfall amount as compared to rainfall estimated from the SCSMEX sounding network. The model yields a slightly higher total convective rain/stratiform rain ratio than the TMI and PR observations. The GCE rainfall spectrum exhibits a greater contribution from heavy rains than those estimated from PR or TMI observations. In addition, the GCE simulations produce much greater amounts of snow and graupel than the TRMM retrievals. The model’s precipitation efficiency of convective rain is close to the observations, but the precipitation efficiency of stratiform rain is much lower than the observations due to large amounts of slowly falling snow and graupel simulated. Compared to observations, the GCE produces compact areas of intensive convection and less anvil cloud, which are consistent with a smaller total cloud fraction and larger domain-averaged OLR.