The 94 GHz profiling radar onboard CloudSat is providing new estimates of vertically resolved upper-tropospheric ice water content on a global scale as part of the A-Train constellation of Earth observing satellite sensors. These estimates are important for understanding the amount, coverage, and structure of ice clouds and are being used as an evaluation tool for representations of upper-tropospheric ice clouds in atmospheric models. This study explores the use of CloudSat retrievals of ice particle size distribution parameters for partitioning the total ice content into ice mass from small and large particles (e.g., clouds versus precipitating hydrometeors) for comparison with global models. Partitioned CloudSat IWC estimates compare favorably with MLS ice content and cloud ice analyses from ECMWF. The partitioned CloudSat ice content is compared to a global atmospheric model that uses cloud-system resolving cloud microphysics within a GCM framework. The results of these analyses illuminate significant differences between the distributions of ice mass versus size as estimated by CloudSat in comparison to its representation in the model.