The myxovirus resistance gene (Mx) expresses antiviral activity in many species, e.g. mouse, human and chicken. It is not clear if the antiviral activity of Mx has evolved in these species to inhibit a set of species-specific pathogens, nor what factors drive Mx evolution in different animal lineages. Therefore, it is important to determine the evolutionary pattern of Mx and positively selected sites which affect the antiviral activity of the Mx gene in mammals and birds. We used sequence comparisons among species to detect positively selected sites by conducting phylogenetic analysis. The two-ratio model was significantly better than the one-ratio model in four species (mouse, rat, chicken and duck, p<0.05). Although selection pressure varied among different lineages, Mx had strong purifying selection in mammals and positive selection in chicken and duck lineages. Relative rate test revealed that Mx evolved faster in chickens than in ducks (Tajima's relative rate test, chi(2)=7.17, p<0.01). In the further analysis using a branch-site model A test, 8 sites were positively selected in the chicken lineage while no positive selection signals were observed for any site in the other lineages. The branch-site model A test had a omega value of 4.374 for the chicken lineage (2Deltal=14.20, d.f.=1, p<0.001). Comparisons of all currently available Mx mRNA sequences showed that these predicted positively selected sites had been fixed in the chicken lineage, suggesting that the chicken Mx gene evolved within the species to resist newly challenging environments. There is an increased selection constraint leading to mammals, while positive selection has acted on the chicken Mx.