HIV-1 strains with a syncytium-inducing phenotype that use CXCR4 (X4 strains) have been associated with faster disease progression and AIDS. Antiviral agents designed to block CXCR4 may prevent the emergence of X4 HIV strains but resistant strains that maintain the X4 phenotype can be raised by sequential passage in cell cultures. We have demonstrated that a laboratory adapted strain (NL4-3) and a cloned clinical isolate (CI-1) of HIV-1 cultured in the presence of the CXCR4 antagonist, AMD3100, became resistant to the compound without a change in co-receptor use. These strains became resistant through divergence with respect to the wild-type virus. Conversely, a clinical isolate made resistant to AMD3100 switched co-receptor use from X4 to R5 through a change in diversity from the original virus population. When dual infection competition/heteroduplex tracking assays were performed, all AMD3100-resistant strains, regardless of co-receptor use showed a significantly diminished fitness compared with the wild-type virus. Single virus infections, at a similar multiplicity of infection, also indicated that the wild-type strains possess better replicative ability than their corresponding resistant strains. Thus, viral resistance development to a CXCR4 antagonist such as AMD3100 is associated with reduced viral fitness.