A statistically significant relationship is found between the seasonal volume October--March of the Diamante River (central Andes mountains of Argentina) and the sea surface temperature (SST) in the Ni¿o3 region of the Pacific Ocean (5 ¿S--5 ¿N, 90 ¿W--150 ¿W), during the previous March--April and simultaneous November--December. The Ni¿o3 SST anomalies are positively correlated with seasonal volume anomalies. A multiple linear regression model for October--March water volume predictions is developed, making use of Ni¿o3 anomalies observed during March--April and six-month November--December Ni¿o3 forecast produced in May. A cross-validation analysis of three-category (tercile) forecasts for the period 1980--1994 reveals a 73% model skill, equal to that of the snowpack thickness model in use. The results of this study are of practical application to hydrologic forecast for water resources management and hydroelectricity generation. This methodology would allow the water resources administrator to have available every year in May a forecast of the water available in the system during the following melting period October--March, before the wintertime snowfalls. ¿ 1999 American Geophysical Union