Given the importance of clouds to climate feedbacks, we assess their simulation quality in climate models using output of satellite simulators. We use satellite simulators to reduce the effects of observational limitations in order to increase the chances that differences between the models and observations represent actual model deficiencies. The assessment is done for both CMIP3 and CMIP5 models to demonstrate if climate model simulations of clouds are improving. The impact of precipitation distribution assumption on the simulated vertical structure of cloud systems is also examined using CAM5. The ability of CMIP3 and CMIP5 models to simulate the climatological distribution of clouds is examined with the International Satellite Cloud Climatology Project (ISCCP) observations. Metrics are computed for the seasonal cycle and geographic distributions of cloud cover as a function of minimum cloud optical thickness, as well as cloud properties stratified by vertical level (low vs. high) and region (tropics versus extratropics). In CMIP3, the simulated clouds covered too little area but were optically too thick. Other important deficiencies included underestimates of subtropical marine boundary layer clouds and middle level cloudiness. Our results indicate small progress to reduce these systematic model biases in statistics of total, low, or high-level cloud cover in going from CMIP3 to CMIP5 models, although progress in individual models can be found in some cases. However, significant progress is found in the reduced amount of very optically thick cloud (tau > 23), particularly over the mid-latitude storm tracks. With increased amounts of clouds with lesser reflectivity, the compensating errors that permit models to simulate the time-mean radiation balance have been reduced. Sensitivity tests are performed with CAM5 to examine the impact of uncertainties from its sub-column precipitation distribution assumption in the radar simulator. We evaluate the vertical distribution of clouds and precipitation simulated in CAM5 over several important cloud regimes and estimate the effects of simulator uncertainties from precipitation sub-column distribution. Preliminary results show that, through an appropriate treatment of uncertainty, the origin of model error can be determined more confidently.