The mouse MHC class I-b molecule H2-M3 has unique specificity for N-formyl peptides, derived from bacteria (and mitochondria), and is thus a pathogen-associated molecular pattern recognition receptor (PRR). To test whether M3 was selected for this PRR function, we studied M3 sequences from diverse murid species of murine genera Mus, Rattus, Apodemus, Diplothrix, Hybomys, Mastomys, and Tokudaia and of sigmodontine genera Sigmodon and Peromyscus. We found that M3 is highly conserved, and the 10 residues coordinating the N-formyl group are almost invariant. The ratio of nonsynonymous and synonymous substitution rates suggests the Ag recognition site of M3, unlike the Ag recognition site of class I-a molecules, is under strong negative (purifying) selection and has been for at least 50-65 million years. Consistent with this, M3 α1α2 domains from Rattus norvegicus and Sigmodon hispidus and from the "null" allele H2-M3b specifically bound N-formyl peptides. The pattern of nucleotide substitution in M3 suggests M3 arose rapidly from murid I-a precursors by an evolutionary leap ("saltation"), perhaps involving intense selective pressure from bacterial pathogens. Alternatively, M3 arose more slowly but prior to the radiation of eutherian (placental) mammals. Older dates for the emergence of M3, and the accepted antiquity of CD1, suggest that primordial class I MHC molecules could have evolved originally as monomorphic PRR, presenting pathogen-associated molecular patterns. Such MHC PRR molecules could have been preadaptations for the evolution of acquired immunity during the early vertebrate radiation.