APOBEC3 deaminases (A3s) provide mammals with an anti-retroviral barrier by catalyzing dC-to-dU deamination on viral ssDNA. Within primates, A3s have undergone a complex evolution via gene duplications, fusions, arms race, and selection. Human APOBEC3C (hA3C) efficiently restricts the replication of viral infectivity factor (vif)-deficient Simian immunodeficiency virus (SIVΔvif), but for unknown reasons, it inhibits HIV-1Δvif only weakly. In catarrhines (Old World monkeys and apes), the A3C loop 1 displays the conserved amino acid pair WE, while the corresponding consensus sequence in A3F and A3D is the largely divergent pair RK, which is also the inferred ancestral sequence for the last common ancestor of A3C and of the C-terminal domains of A3D and A3F in primates. Here, we report that modifying the WE residues in hA3C loop 1 to RK leads to stronger interactions with substrate ssDNA, facilitating catalytic function, which results in a drastic increase in both deamination activity and in the ability to restrict HIV-1 and LINE-1 replication. Conversely, the modification hA3F_WE resulted only in a marginal decrease in HIV-1Δvif inhibition. We propose that the two series of ancestral gene duplications that generated A3C, A3D-CTD and A3F-CTD allowed neo/subfunctionalization: A3F-CTD maintained the ancestral RK residues in loop 1, while diversifying selection resulted in the RK → WE modification in Old World anthropoids' A3C, possibly allowing for novel substrate specificity and function.