Abstract Background Allorecognition, the ability of an organism to distinguish self from non-self, occurs throughout the entire tree of life. Despite the prevalence and importance of allorecognition systems, the genetic basis of allorecognition has rarely been characterized outside the well-known MHC (Major Histocompatibility Complex) in vertebrates and SI (Self-Incompatibility) in plants. Where loci have been identified, their evolutionary history is an open question. We have previously identified the genes involved in self/non-self recognition in the colonial ascidian Botryllus schlosseri, and we can now begin to investigate their evolution. In B. schlosseri, colonies sharing 1 or more alleles of a gene called FuHC (Fusion Histocompatibility) will fuse. Protein products of a locus called fester, located ~300 kb from FuHC, have been shown to play multiple roles in the histocompatibility reaction, as activating and/or inhibitory receptors. We test whether the proteins encoded by this locus are evolving neutrally or are experiencing balancing, directional, or purifying selection. Results Nearly all of the variation in the fester locus resides within populations. The 13 housekeeping genes (12 nuclear genes and mitochondrial cytochrome oxidase I) have substantially more structure among populations within groups and among groups than fester. All polymorphism statistics (Tajima's D, Fu and Li's D* and F*) are significantly negative for the East Coast A-type alleles, and Fu and Li's F* statistic is significantly negative for the West Coast A-type alleles. These results are likely due to selection rather than demography, given that 10 of the housekeeping loci have no populations with significant values for any of the polymorphism statistics. The majority of codons in the fester proteins have ω values < 1, but 15–27 codons have > 95% posterior probability of ω values > 1. Conclusion Fester proteins are evolving non-neutrally. The polymorphism statistics are consistent with either purifying selection or directional selection. The ω statistics show that the majority of the protein is experiencing purifying selection (ω < 1), but that 15–27 codons are undergoing either balancing or directional selection: ω > 1 is compatible with either scenario. The distribution of variation within and among populations points towards balancing selection and away from directional selection. While these data do not provide unambiguous support for a specific type of selection, they contribute to our evolutionary understanding of a critical biological process by determining the forces that affect loci involved in allorecognition.