The replication fitness of HIV-1 drug-resistant mutants has been measured using either multiple-cycle or single-cycle assays (MCAs or SCAs); these assays have not been systematically compared. We developed an MCA and an SCA that utilized either intact or env-deleted recombinant viral vectors, respectively, in which virus-infected cells were detected by flow cytometry of a reporter gene product. Fitness was measured using each assay for 11 protease mutants, 9 reverse transcriptase mutants, and two mutants with mutations in gag p6, which is important for the release of virus particles from the cell membrane. In the SCA, fitness (replication capacity [RC]) was defined as the proportion of cells infected by the mutant compared to the wild type 40 h after infection. MCA fitness (1+s) was determined by comparing the changes in the relative proportions of cells infected by the mutant and the wild type between 3 and 5 days after infection. Five protease mutants showed statistically different fitness values by the MCA versus the SCA: the D30N, G48V, I50V, I54L, and I54M mutants. When all the mutants were ranked in order from most to least fit for both assays, 4 protease mutants moved more than 5 positions in rank: the D30N, I54L, I54M, and V82A mutants. There were no significant differences in fitness for the gag p6 or reverse transcriptase mutants. We propose that discordant results in the MCA and SCA are due to alterations in late events in the virus life cycle that are not captured in an SCA, such as burst size, cell-to-cell transmission, or infected-cell life span.