Coreceptor specificity of human immunodeficiency virus type 1 (HIV-1) strains is generally defined in vitro in cell lines expressing CCR5 or CXCR4, but lymphocytes and macrophages are the principal targets in vivo. CCR5-using (R5) variants dominate early in infection, but strains that use CXCR4 emerge later in a substantial minority of subjects. Many or most CXCR4-using variants can use both CXCR4 and CCR5 (R5X4), but the pathways that are actually used to cause infection in primary cells and in vivo are unknown. We examined several R5X4 prototype and primary isolates and found that they all were largely or completely restricted to CXCR4-mediated entry in primary lymphocytes, even though lymphocytes are permissive for CCR5-mediated entry by R5 strains. In contrast, in primary macrophages R5X4 isolates used both CCR5 and CXCR4. The R5X4 strains were also more sensitive than R5 strains to CCR5 blocking, suggesting that interactions between the R5X4 strains and CCR5 are less efficient. These results indicate that coreceptor phenotyping in transformed cells does not necessarily predict utilization in primary cells, that variability exists among HIV-1 isolates in the ability to use CCR5 expressed on lymphocytes, and that many or most strains characterized as R5X4 are functionally X4 in primary lymphocytes. Less efficient interactions between R5X4 strains and CCR5 may be responsible for the inability to use CCR5 on lymphocytes, which express relatively low CCR5 levels. Since isolates that acquire CXCR4 utilization retain the capacity to use CCR5 on macrophages despite their inability to use it on lymphocytes, these results also raise the possibility that a CCR5-mediated macrophage reservoir is required for sustained infection in vivo.