In aviation, there is always a need for better weather forecast products. Whether it is higher resolution models in the spatial or temporal scale or improved forecast algorithms, any advancement made in current forecast products are beneficial to air traffic management operations and research. This fact was especially apparent while developing a probabilistic airport capacity model for use in traffic management decision support tools. The model predicts the airport arrival rate (AAR) from weather forecast products that are primarily used for the terminal area: the Localized Aviation MOS Product (LAMP) and the Terminal Aerodrome Forecast (TAF). The initial version of this model was trained to forecast capacity at EWR, but we have expanded our work to include ORD in recent months and have included weather data from 2008 through 2010. For both airports, the model has shown skill in predicting the AAR; however it has been found that the inaccuracy and incompleteness of the model can be partially traced back to the weather forecast products that are being utilized. Both of the products are widely used by operational meteorologists and dispatchers and each of them has their benefits, but it appears that enhancements are needed. These include: improved probabilistic weather forecasts, higher resolution thunderstorm probability forecasts both spatially and temporally, better algorithms that handle mesoscale phenomena around each airport, and higher temporal resolution forecasts in the terminal area. Each one of these improvements would make the integration of weather and air traffic management much smoother and potentially make the outcomes of decision support tools more accurate. In this paper we present an overview of our airport capacity estimation model, the weather products that we are using to run our model, problems encountered while incorporating the data into our model, and potential improvements to weather products that would be useful for future TFM work.