Homogenization of duplicated genes is an important factor for gene family evolution. In the previous study, we developed a method, named 4-2-4 here, to detect partial homogenization with high sensitivity and high specificity using quartets. A quartet is a set of four genes generated by a duplication event and the subsequent speciation of two closely related species. We searched the human and macaque genomes and found 430 nonredundant quartets, which correspond to primate-specific paralogs. The prevalence of homogenization in these quartets was 10.0% (43/430), which was ca. one-third of that (29.8% = 206/691) in the rodent-specific nonredundant quartets obtained through comparison of mouse and rat genomes. Part of this difference comes from the fact that primate paralogs tend to be more remotely located to each other than rodent paralogs, and the remainder may be explained by the inherent difference in the neutral evolutionary rate between the primate and rodent lineages. A statistical analysis taking account of the effects of false negatives uncovered negative correlations between sequence divergence and homogenization prevalence both in primates and rodents. Further statistical analyses controlling for false-negative rates and sequence divergences revealed two characteristics shared by the primate and rodent paralogs; 1) significant negative correlations of the homogenization prevalence with physical distances, and 2) no significant correlation between the prevalence and relative transcriptional orientations. Patterns of the homogenization in the genomic alignments of human-macaque quartets indicate that gene conversion, rather than unequal crossing-over, is the major cause of the homogenization.