The Scripps Experimental Climate Prediction Center (ECPC) has been making experimental, near real-time seasonal global forecasts since Sept. 26, 1997 with the NCEP global spectral model used for the reanalysis. Images of these forecasts, at daily to seasonal time scales, are provided on the world wide web (WWW) and digital forecast products are provided on the ECPC anonymous ftp site to interested researchers. These forecasts are increasingly being used to drive regional models at the ECPC and elsewhere as well as various application models, including a fire severity model. The purpose of this talk is to describe the global to regional forecast and analysis system, various biases and errors in the forecasts, as well as the significant skill of the forecasts. Forecast near-surface meteorological parameters, including temperature, precipitation, soil moisture, relative humidity, wind speed, and a fire weather index (FWI; a nonlinear combination of temperature, windspeed, and relative humidity) are skillful at weekly to seasonal time scales over much of the US and for many global regions. These experimental results suggest there is substantial forecast skill, out to at least a season, to be realized from current dynamical models. These results also suggest that fire severity forecasts will show corresponding skill, especially since severity forecasts include low frequency vegetation influences.