The development of an effective vaccine against human immunodeficiency virus is considered to be the most practicable means of controlling the advancing global AIDS epidemic. Studies with the domestic cat have demonstrated that vaccinal immunity to infection can be induced against feline immunodeficiency virus (FIV); however, protection is largely restricted to laboratory strains of FIV and does not extend to primary strains of the virus. We compared the pathogenicity of two prototypic vaccine challenge strains of FIV derived from molecular clones; the laboratory strain PET(F14) and the primary strain GL8(414). PET(F14) established a low viral load and had no effect on CD4(+)- or CD8(+)-lymphocyte subsets. In contrast, GL8(414) established a high viral load and induced a significant reduction in the ratio of CD4(+) to CD8(+) lymphocytes by 15 weeks postinfection, suggesting that PET(F14) may be a low-virulence-challenge virus. However, during long-term monitoring of the PET(F14)-infected cats, we observed the emergence of variant viruses in two of three cats. Concomitant with the appearance of the variant viruses, designated 627(W135) and 628(W135,) we observed an expansion of CD8(+)-lymphocyte subpopulations expressing reduced CD8 beta-chain, a phenotype consistent with activation. The variant viruses both carried mutations that reduced the net charge of the V3 loop (K409Q and K409E), giving rise to a reduced ability of the Env proteins to both induce fusion and to establish productive infection in CXCR4-expressing cells. Further, following subsequent challenge of naïve cats with the mutant viruses, the viruses established higher viral loads and induced more marked alterations in CD8(+)-lymphocyte subpopulations than did the parent F14 strain of virus, suggesting that the E409K mutation in the PET(F14) strain contributes to the attenuation of the virus.