Natural selection is expected to leave distinctive signatures on patterns of neutral variation that are tightly linked to a site carrying an advantageous mutation. This notion is the basis for molecular population genetics approaches to the analysis of adaptations in the human genome. In principle, this is an attractive prospect; in practice, several complications and challenges make it difficult to draw unambiguous conclusions about natural selection. In this article, we discuss some general issues on the studies detecting the signature of natural selection in light of a recent paper on the prion protein gene (PRNP). This study hypothesized that PRNP evolved under balancing selection as a result of widespread cannibalistic practices.