To meet these critical needs of high sensitivity and specificity, which are not adequately addressed by the available methods in the field21–23, we have developed a somatic point mutation caller, MuTect. During its development MuTect was used in numerous studies1–4,7,19,24–35. Here we describe the final and publicly available version of MuTect including the rationale behind its different components. We also estimate its performance as a function of the aforementioned factors using benchmarking approaches that, to our knowledge, have not been described before; through independent experimental validation in previous studies3,4,7,19,24–30; as well as by applying our method to datasets analyzed in other publications21,36,37. We demonstrate that our method is several times more sensitive than other methods for events at low allelic fractions while remaining highly specific, allowing for deeper exploration of the mutational landscape of highly impure tumor samples and the subclonal evolution of tumors.