Although several models have been developed to describe the evaporative casting of dense polymer films, none of them has included the convective transport terms which arise owing to the densification which occurs. In this paper we first describe a new finite element solution to the binary nonisothermal evaporative casting process which is used to confirm the predictions of the finite difference solution recently developed by Shojaie et al. (J. Mater. Process. Manu. Sci., 1 (1992) 181). This comparison then establishes that systematic deviations between experimental measurements and the model predictions of Shojaie et al. do not arise because of any inaccuracies in the numerical solution methodology. We then explore whether these deviations arise owing to omission of the convective transport terms by Shojaie et al. and all other modeling efforts. We employ a scaling analysis to demonstrate that convective transport effects can be significant in the evaporative casting process. The convective transport terms then are incorporated into the nonisothermal evaporative casting model of Shojaie et al. for the cellulose acetate/acetone system. The model predictions both with and without convective transport are compared to real-time gravimetric and surface-temperature data. This study indicates that ignoring convective transport can result in differences as large as 40% in the model predictions. The model predictions incorporating convective transport are in quantitative agreement with real-time data at short evaporation times, but progressively deviate at longer times. This deviation is thought to be due to the use of an equilibrium equation-of-state for the solution density as a function of concentration.