A method for seasonal predictions of U.S. precipitation based on tropical Pacific sea surface temperature (SST) anomalies is developed in this letter with the singular value decomposition (SVD). This method was applied to the 1997/98 El Ni¿o. The 1997 summer and 1997/98 winter precipitation over the United States were predicted using forecast SST from the National Centers for Environmental Prediction (NCEP) coupled ocean-atmosphere model. A cross validation, based on the percentages of hit between the hindcast and observed precipitation for the past El Ni¿o and La Ni¿a events, indicates a certain degree of predictability in both winter and summer seasons. During El Ni¿o, significant predictability of summer precipitation is found over the Northern Plains and Atlantic States, while during La Ni¿a it is in the Midwest. Above normal precipitation in the Northern Plains and the Midwest and below normal precipitation in the Atlantic States is associated with a warm phase of the El Ni¿o-Southern Oscillation (ENSO) in summer. For winter precipitation, significant predictability is detected over the Gulf Coast States, the Southern Plains and California. Over these regions, wetter conditions are generally associated with the warm phase of ENSO. ¿ 1999 American Geophysical Union