Accurate measurements of rainfall are necessary for understanding the hydrologic cycle. Current rainfall retrieval algorithms rely on a database generated using a cloud resolving model, where the accuracy of the database depends on the accuracy of the model. At present it is not known to what degree the retrieval algorithms are dependent upon the accuracy of the cloud models. To answer this question, two models, the Goddard Cumulus Ensemble model and the Regional Atmospheric Modeling System (RAMS) at Colorado State University, were used to generate databases for a rainfall retrieval algorithm. Multiple model runs with varying microphysical parameterizations are used to examine this problem. The retrieval algorithm uses the rms difference in computed and observed brightness temperatures to retrieve the surface rainfall, column water and column ice. By running the retrieval algorithm using the various databases and examining how the retrieved rainfall, column water and column ice vary, the sensitivity of the retrieval algorithm to the model-generated database can be quantified.