When the cost of hedging is nil, the conditional capital asset pricing model (CAPM) holds. We empirically test the conditional CAPM by regressing asset returns onto the product of their conditional betas and market returns. Estimated intercepts are not statistically different from zero, implying that the conditional CAPM successfully explains the conditional level of asset returns. Yet, unconditional betas do not explain the cross section of average asset returns; the unconditional CAPM fails. We show why and how the success of the conditional CAPM actually explains the failure of the unconditional CAPM, thereby rationalizing the coexistence of these two intriguing results. This paper was accepted by Gustavo Manso, finance. Funding: The University of Texas at Dallas and the University of Toronto provided financial support. Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2022.4381 .