In order to more realistically represent both radiative and microphysical processes in anvil clouds in GCMs, the cloud fraction due to anvil clouds should be included by representing, in a simplified fashion, the physical processes that form, maintain, and dissipate anvil clouds. With the recent widespread adoption of prognostic cloud water/ice schemes in GCMs, anvil cloud formation is represented as a source term due to detrainment from deep cumulus convection. Once formed, an anvil or cirrus cloud is subject to many subgrid-scale, non-convective processes, including condensation/deposition due to ascent, radiative cooling, mesoscale circulations, and turbulent mixing, that tend to increase its area and ice content, and to precipitation and sublimation that tend to decrease its area and ice content. The fraction of a grid cell occupied by anvil clouds is largely determined by the history of the anvil clouds, so that a prognostic cloud fraction parameterization is appropriate. Such an approach has been developed by Tiedtke (1993) and extended by Randall and Fowler (1999), but has not been examined using cloud-resolving models (CRMs) or tested against observations except indirectly using global, monthly averaged datasets. We are using 2D and 3D CRM simulations of the life cycle of anvil clouds to examine in detail the physical processes that determine the cloud fraction of anvil clouds. We will present our results at the symposium.