This article proposes a mean-variance optimization and portfolio frontier analysis of energy risk management with carbon assets, introduced in January 2005 as part of the EU Emissions Trading Scheme. In a stylized exercise, we compute returns, standard deviations and correlations for various asset classes from April 2005 to January 2009. Our central result features an expected return of 3% with a standard deviation < 0.06 by introducing carbon assets -carbon futures and CERs-in a diversified portfolio composed of energy (oil, gas, coal), weather, bond, equity risky assets, and of a riskless asset (U.S. T-bills). Besides, we investigate the characteristics of each asset class with respect to the alpha, beta, and sigma in the spirit of the CAPM. These results reveal that carbon, gas, coal and bond assets share the best properties for composing an optimal portfolio. Collectively, these results illustrate the benefits of carbon assets for diversification purposes in portfolio management, as the carbon market constitutes a segmented commodity market with specific risk factors linked to the EU Commission's decisions and the power producers' fuel-switching behavior.JEL codes: C61; G11; Q40