ABCG subfamily proteins are highly enriched in terrestrial plants. Many of these proteins secrete secondary metabolites that repel or inhibit pathogens. To establish why the ABCG subfamily proteins proliferated extensively during evolution, we constructed phylogenetic trees from a broad range of eukaryotic organisms. ABCG proteins were massively duplicated in land plants and in oomycetes, a group of agronomically important plant pathogens, which prompted us to hypothesize that plant and pathogen ABCGs coevolved. Supporting this hypothesis, full-size ABCGs in host plants (Arabidopsis thaliana and Glycine max) and their pathogens (Hyaloperonospora arabidopsidis and Phytophthora sojae, respectively) had similar divergence times and patterns. Furthermore, generalist pathogens with broad ranges of host plants have diversified more ABCGs than their specialist counterparts. The hypothesis was further tested using an example pair of ABCGs that first diverged during multiplication in a host plant and its pathogen: AtABCG31 of A. thaliana and HpaP802307 of H. arabidopsidis. AtABCG31 expression was activated following infection with H. arabidopsidis, and disrupting AtABCG31 led to increased susceptibility to H. arabidopsidis. Together, our results suggest that ABCG genes in plants and their oomycete pathogens coevolved in an arms race, to extrude secondary metabolites involved in the plant's defense response against pathogens.