This study applies financial portfolio theory to determine efficient electricity-generating technology portfolios for the United States and Switzerland, adopting an investor point of view. Expected returns are defined by the rate of decrease of power generation cost (with external costs included), their volatility, by its standard deviation. The 2003 portfolio contains Coal, Nuclear, Gas, Oil, and Wind in the case of the United States, and Nuclear, Storage hydro, Run of river, and Solar in the case of Switzerland, a country without domestic supplies of fossil fuels. Since shocks in generation costs are found to be correlated, Seemingly Unrelated Regression Estimation (SURE) is used to filter out the systematic component of the covariance matrix of the cost changes. Results suggest that as of 2003, the feasible maximum expected return (MER) electricity portfolio for the United States contains more Coal, Nuclear, and Wind than actual but markedly less Gas and Oil. By way of contrast, the minimum variance (MV) portfolio combines markedly more Oil, Coal, Nuclear, and Wind but almost no Gas. Therefore, regardless of the choice between MER and MV, U.S. utilities as investors are substantially inside the efficient frontier. This is even more true of their Swiss counterparts, likely due to continuing regulation of electricity markets. -- efficiency frontier ; energy ; electricity ; portfolio theory ; Seemingly Unrelated Regression Estimation (SURE)