We investigate the portfolio choices of mean-variance-optimizing investors who use sample evidence to update prior beliefs centered on either risk-based or characteristic-based pricing models. With dogmatic beliefs in such models and an unconstrained ratio of position size to capital, optimal portfolios can differ across models to economically significant degrees. The differences are substantially reduced by modest uncertainty about the models’ pricing abilities. When the ratio of position size to capital is subject to realistic constraints, the differences in portfolios across models become even less important and are nonexistent in some cases.