The extant literature demonstrates the importance of stock return predictability for portfolio allocation. The usefulness of incorporating return predictability into portfolio decisions is most evident for Bayesian investors who build their portfolios based on their prior beliefs. I show that the magnitude of economic significance of stock return predictability largely depends on the choice of prior beliefs. An investor could suffer substantial utility loss when he delegates portfolio management to a manager with a different belief about stock return predictability. The consideration of Bayesian prior robustness in portfolio analysis can be as important as return predictability itself.