Coval, Jurek and Stafford (2009, hereafter CJS) present a state-contingent framework for pricing CDO, in which they integrate the generalized form of Merton’s (1974) structural model with the CAPM. We demonstrate that this seemingly intuitive model violates a risk-neutral valuation relationship, and therefore results in biased state-contingent payoffs of the firm’s asset. We suggest a correct and generalized model, and revisit the pricing of CDO. We argue that this modification, though not translating into any significant difference in pricing traded CDOs, enable CJS to be a more accurate and strict, and thus powerful one. Coval, Jurek and Stafford (2009, hereafter CJS) modify Merton’s (1974) model in a way that asset returns are driven by a combination of market shocks, Zm, and idiosyncratic shocks, Zi,ε. In their model, the sensitivity of asset returns to market shocks is assumed to be proportional to its CAPM beta. Also, most important, they leave the distribution of the market shocks unspecified, allowing asset returns to have arbitrary, non-Gaussian distributions. Furthermore, they impose a restriction on excess returns under the name of the Sharpe (1964) and Lintner (1965) CAPM requirement. This seemingly intuitive model of CJS, however, has some drawbacks. In this comment, we point out two of them and suggest a modified model where they can be resolved.