It is generally agreed that effective management of convective weather in congested airspace requires decision support tools that translate the weather products and forecasts into forecasts of ATC impacts and then use those ATC impact forecasts to suggest air traffic management strategies. In future trajectory-based operations, it will be necessary to automatically generate flight trajectories through or around convective weather that pilots will find acceptable. A critical first step, needed in both today's air traffic management environment and in the highly automated systems of the future, is a validated model for airspace that pilots will seek to avoid. At the 12th Conference on Aviation, Range and Aerospace Meteorology (Atlanta, 2006), we reported on an initial Convective Weather Avoidance Model (CWAM1) (DeLaura and Evans; 2006). The CWAM1 outputs are three dimensional deterministic and probabilistic weather avoidance fields (WAFs). CWAM1 used Corridor Integrated Weather System (CIWS) VIL and echo top fields and National Lightning Detection Network (NLDN) data to predict aircraft deviations and penetration. CWAM1 was developed using more than 500 aircraft-convective weather encounters in the Indianapolis Air Route Traffic Control Center (ZID ARTCC) airspace. CWAM1 gave the greatest weight to the difference between flight altitude and the 18 dbZ radar echo top with precipitation intensity playing a secondary role. The deviation prediction error rate in CWAM1 was approximately 25%. This paper presents a new model (CWAM2), based on the analysis of trajectories from several ARTCCs [Indianapolis (ZID), Cleveland (ZOB) and meteorological deviation predictors. Additional weather factors that are considered include vertical storm structure (upper level reflectivity and the height of the VIL centroid derived from the NSSL 3D reflectivity mosaic), vertical and horizontal storm growth, the spatial variation in VIL and echo top fields and storm motion.