Here, a protein known as MX2 is shown to be a major effector of interferon-α-mediated resistance to HIV-1 infection: susceptibility of the HIV-1 virus to inhibition by MX2 is dictated by the Capsid region of the viral Gag protein, and inhibition occurs at a late post-entry step of infection. Two groups report in this issue of Nature that the human interferon-induced GTP-binding protein MX2 is a potent inhibitor of human immunodeficiency virus type 1 (HIV-1) and a number of other lentiviruses. For some years it had been known that the related protein MX1 can inhibit HIV-1 replication in humans, but MX2 was thought to be devoid of antiviral activity. The anti-HIV-1 action of MX2 is much less dependent on GTPase activity than is the broader antiviral activity of MX1, pointing to possible mechanistic differences between them. Animal cells harbour multiple innate effector mechanisms that inhibit virus replication. For the pathogenic retrovirus human immunodeficiency virus type 1 (HIV-1), these include widely expressed restriction factors1, such as APOBEC3 proteins2, TRIM5-α3, BST2 (refs 4, 5) and SAMHD1 (refs 6, 7), as well as additional factors that are stimulated by type 1 interferon (IFN)8,9,10,11,12,13,14. Here we use both ectopic expression and gene-silencing experiments to define the human dynamin-like, IFN-induced myxovirus resistance 2 (MX2, also known as MXB) protein as a potent inhibitor of HIV-1 infection and as a key effector of IFN-α-mediated resistance to HIV-1 infection. MX2 suppresses infection by all HIV-1 strains tested, has equivalent or reduced effects on divergent simian immunodeficiency viruses, and does not inhibit other retroviruses such as murine leukaemia virus. The Capsid region of the viral Gag protein dictates susceptibility to MX2, and the block to infection occurs at a late post-entry step, with both the nuclear accumulation and chromosomal integration of nascent viral complementary DNA suppressed. Finally, human MX1 (also known as MXA), a closely related protein that has long been recognized as a broadly acting inhibitor of RNA and DNA viruses, including the orthomyxovirus influenza A virus15,16, does not affect HIV-1, whereas MX2 is ineffective against influenza virus. MX2 is therefore a cell-autonomous, anti-HIV-1 resistance factor whose purposeful mobilization may represent a new therapeutic approach for the treatment of HIV/AIDS.