Aside from model evaluation results presented as robust highest concentration statistics and quantile-quantile plots, it is important to understand other aspects of the behavior of the models being evaluated. This paper presents results obtained from an analysis of residual plots and conditions associated with the highest several predicted and observed concentration estimates. It indicates how the models that were evaluated compare with observations and each other as a function of stability, distance, wind speed, mixing height, and other important modeling parameters. The results will show areas in which AERMOD has exhibited improvements over existing modeling techniques, and other areas that may need additional study.