The Vif (viral infectivity factor) protein of human immunodeficiency virus–1 (HIV-1) is essential for the production of infectious virus by T lymphocytes and macrophages, the natural targets for HIV-1 infection. Vif is dispensable for HIV infection in certain transformed cell lines (“permissive” cells) but is required for propagative infection in others (“nonpermissive” cells). Vif is thought to inhibit an antiviral pathway recently demonstrated to require the cellular protein CEM 15 (1). This protein, also called APOBEC3G (2), is a member of the family of cytidine deaminases. Because recent studies have suggested that some mammalian cytidine deaminases target single-stranded DNA (3), we tested the hypothesis that Vif may be required to prevent the editing of minusstrand viral DNA synthesized by virions produced in nonpermissive cells. Thus, nonpermissive H9 cells (4) were exposed to virus produced by HeLa cells transfected with wild-type HIV-1 (pNL4-3) or a variant in which a portion of the vif gene had been deleted (vif). The virions produced by H9 cells 24 to 48 hours after infection were recovered. Target cells (CD4, HeLa-derived P4 cells) were then exposed to these virions and incubated for 5 hours at 37 C, and newly reverse transcribed viral DNA was isolated (4). Degenerate primers that would not be affected by the editing of minusstrand DNA by cytidine deaminases were then used to amplify viral sequences in env and U5. The products were then cloned and sequenced. Guanine to adenine (G3A) changes were significantly more frequent in DNA synthesized by vif virions than in DNA produced by wildtype virions (P 0.0001 and P 0.01)(Fig. 1A; fig. S1A and fig. S2). In contrast, no difference in the frequency of G3A changes could be observed in viral DNA synthesized by wild-type and vif viruses produced in permissive cells (P4 cells) under the same conditions, or those generated in HeLa cells (fig. S1). G3A changes in the HIV plus-strand DNA can be directly explained by modifications of cytidine to uridine (C3U) in the minus-strand of DNA by a cytidine deaminase, because U is subsequently read as T by DNA polymerases. Such changes might also be compatible with hypermutation by reverse transcriptase, a phe-