The suitability of using factors or benchmarks to measure portfolio performance is analysed. Fama and French factors are constructed from Russell US stock indexes and then directly utilized as benchmarks. The interpretation of factors as zero-investment benchmarks makes it difficult to explain performance measurement as the comparison of active versus passive management, given the short selling restrictions often applied to mutual funds. Empirical results reveal similar biases in extended Jensen's alphas in models with both factors and with benchmarks and with convexity and nonnegativity restrictions. Selection of the benchmarks has a more important effect than the model type chosen.