A stochastic cloud-radiation model has been shown to do a good job of representing the domain-averaged shortwave fluxes when evaluated using observations. One year of continuously sampled cloud property observations from all three Atmospheric Radiation Measurement (ARM) Program's Clouds and Radiation Testbed (CART) sites (Southern Great Plains, Tropical Western Pacific, and North Slopes of Alaska) are analyzed and then input into the model on an hourly basis. We compare the output from the multiple-layer stochastic model to that of the single-layer version of the model previously used. Both will also be compared to plane parallel model output and independent observations. These results will form the basis for a shortwave cloud-radiation parameterization that will incorporate the influence of the stochastic approach on the calculated radiative fluxes. Initial results using this resulting parameterization in a single-column model will be shown.