This paper describes a system for preparing monthly outlooks for fire-weather elements in the United States. The system is based on multiple regression equations that specify monthly mean anomalies of precipitation, temperature, dewpoint, and wind speed from concurrent anomalies of the 700 mb height field plus local persistence. The equations are derived by applying a forward selection (screening) procedure with cutoff criteria determined by Monte Carlo simulations and physical reasoning. Averaged over all months, the equations specify 68% (72%) of the temperature variance in the contiguous United States (Alaska) by means of only four variables. The corresponding statistics for dewpoint are 57% (65%) and 3 terms, precipitation 41% (43%) and 3 terms, and wind speed 37% (32%) and 2 terms. The equations were derived from 19 or 20 years of observed 700 mb heights by assuming a perfect circulation forecast. In practice, however, they are applied to prognostic heights prepared monthly by the NOAA Climate Analysis Center and therefore suffer considerable loss of skill. We are now evaluating this loss by a detailed verification study that will allow the outlooks to be expressed in probability form. © by the Society of American Foresters. Abstract reproduced by permission.