Entry of human immunodeficiency virus type 1 (HIV-1) into target cells is mediated by binding of the surface envelope glycoprotein to the CD4 molecule. Interaction of the resulting CD4-glycoprotein complex with alpha- or beta-chemokine receptors, depending on the biological phenotype of the virus, then initiates the fusion process. Here, we show that primary HIV-2 isolates and biological clones, in contrast to those of HIV-1, may use a broad range of coreceptors, including CCR-1, CCR-3, CCR-5, and CXCR-4. The syncytium-inducing capacity of these viruses did not correlate with the ability to infect via CXCR-4 or any other coreceptor. One cell-free passage of the intermediate isolates in mitogen-stimulated, CD8+ cell-depleted peripheral blood mononuclear cells resulted in the outgrowth of variants with CCR-5 only, whereas the coreceptor usage of late and early isolates did not change. Since HIV-2 is less pathogenic in vivo than HIV-1, these data suggest that HIV pathogenicity in vivo is not directly related to the spectrum of coreceptors used in in vitro systems.