BACKGROUND: We compared two strategies for treating patients infected with multidrug-resistant. human immunodeficiency virus (HIV). METHODS: Patients with multidrug-resistant HIV and HIV RNA levels of more than 5000 copies per milliliter were randomly assigned to a four-month structured interruption of treatment followed by a change in antiretroviral regimen (treatment-interruption group) or to an immediate change in regimen (control group). Genotypic and phenotypic resistance testing was performed. Disease progression, death, and changes in genotypic resistance, CD4 cell counts, HIV RNA levels, and quality of life were assessed. RESULTS: After a median follow-up of 11.6 months, disease progression or death occurred in 22 of the 138 patients in the treatment-interruption group and in 12 of the 132 patients in the control group (P=0.01), with a hazard ratio of 2.57 (95 percent confidence interval, 1.2 to 5.5) for the treatment-interruption group. There were eight deaths in each group. In the treatment-interruption group, the mutant HIV populations completely or partially reverted to wild type by four months in 64.0 percent of patients. As compared with the control group, the treatment-interruption group had a mean CD4 cell count that was 85 cells per cubic millimeter lower from months 0 through 4 (P<0.001), 47 cells per cubic millimeter lower from months 5 through 8 (P<0.001), and 31 cells per cubic millimeter lower after eight months (P=0.11). The mean HIV RNA levels were 1.2 log copies per milliliter higher (on a base-10 scale) in the treatment-interruption group during months 0 through 4 (P<0.001), but they were not significantly different from those in the control group after month 4. The overall quality of life was similar in the two groups. CONCLUSIONS: In patients infected with multidrug-resistant HIV, structured interruption of treatment was associated with greater progression of disease and did not confer immunologic or virologic benefits or improve the overall quality of life.