This article studies the performance of US actively managed mutual funds with traditional benchmarks and find, as previous studies, that they obtain negative performance. We believe that this pessimistic result is explained by the excessively high expenses charged by managers or by imperfect measures of expenses. We propose to separate the issue of how much value the manager creates from the issue of how much expenses she charges. We hence propose a performance model in which a benchmark is adjusted for information risk and find, on average, positive and significant performance.