Understanding the nature of MHC class I presentation preferences is a challenging prospect. Large sets of peptide–MHC-class I complexes have been screened for their binding affinities and recent studies have shown that HLA-A share a preference for binding peptides derived from pathogens; however, no mechanism explaining the observed preferences has been demonstrated so far. In this issue of the European Journal of Immunology, a study demonstrates that HLA-A, but not HLA-B, preferentially recognises peptides enriched in amino acids encoded by sequences with low G+C content, and therefore recognises peptides associated with pathogens – low G+C content being a general feature of lower organisms. The authors of the study provide exciting results contributing to the understanding of the nature of MHC-I presentation preferences and MHC-I evolution. Although significant results are presented by the authors, here, we challenge the interpretation whereby HLA-A has been evolutionarily selected for such a function and appeal for the use of comparative phylogenetic methods to substantiate it. We propose a method to ascertain whether ancestral MHC recognised peptides from pathogens and hence was involved in the non-specific recognition of such organisms. Moreover, we suggest that ancestral MHC may have been involved in innate immune responses before being recruited for adaptive immunity