This paper describes a class of weather-weighted periodic auto-regressive sector demand prediction models. The periodic auto-regressive model captures both the mid-term (30 minutes to 2 hours) trend based on the historical data, and the short-term (less than 30 minutes) transient response based on recent observations. For severe weather days, the model uses the three-dimensional weather information, considering both storm locations and echo tops, to form a weather factor to adjust the predictions. Unlike the traditional trajectory-based sector demand prediction methods, which predict the behavior of the National Airspace System adequately for short durations of up to 20 minutes and are vulnerable to the weather uncertainties, this class of models provides reliable short to mid-term sector demand predictions which account for the weather uncertainty.