Abstract: In this paper we investigate whether there are any significant differences in the ability of constant and time-varying expected return asset pricing models to detect superior performance in hedge funds. Our results strongly suggest that the static models traditionally employed to measure and evaluate hedge fund performance are misspecified. Allowing for conditioning information to predict changes in the risk and performance measures of hedge funds increases the statistical significance of the performance evaluation. In addition, incorrectly assuming constant expected returns appears to lead to underestimation of the abnormal performance of hedge funds.