Successfully predicting the phase and strength of the dominant winter mode of variability for the Northern Hemisphere (NAO/AO) is considered the most important future breakthrough in winter climate prediction; however skillful prediction of the index has been elusive. In this Letter we present a snow index constructed from observed summer and fall anomalies that is more highly correlated than the observed value of the winter AO with winter surface temperatures in the eastern United States. Ease of use and the potential for greater predictive skill could potentially render forecasting the phase and strength of the NAO/AO irrelevant as a tool for prediction of U. S. winter surface temperatures.