This paper uses financial portfolio theory to demonstrate how the energy mix consumed in the United States could be chosen given a national goal to reduce the risks to the domestic macroeconomy of unanticipated energy price shocks. An efficient portfolio frontier of U.S. energy consumption is constructed using time-varying variances and covariances estimated with generalized autoregressive conditional heteroskedastic models. The set of efficient portfolios developed are intended to minimize the impact of price shocks, but are not the least cost energy consumption bundles. The results indicate I-hat while the electric utility industry is operating close to the minimum variance position, a shift towards coal consumption would reduce price volatility for overall U.S. energy consumption. With the inclusion of potential externality costs, the shift remains away from oil but towards natural gas instead of coal.