Most individuals with multidrug-resistant HIV who switch to a new therapeutic regimen containing a single fully effective agent experience incomplete viral suppression. We postulated that interruption of antiretroviral therapy prior to the introduction of such a regimen would improve long-term virological outcomes. Thirty, three-class experienced, enfuvirtide-naive individuals with detectable drug-resistant viraemia were randomized to an immediate enfuvirtide/optimized-background treatment regimen or a 16-week treatment interruption followed by enfuvirtide/optimized-background treatment regimen. The median CD4+ T-cell count and viral load at study entry were 39 cells/mm and 4.72 log10 copies RNA/ml, respectively. There was no evidence of any virological or immunological benefit associated with the interruption. In multivariate analysis, only the baseline phenotypic susceptibility score was predictive of treatment response at week 48 (P=0.009). Only 40% of individuals had evidence of a shift in drug-resistance genotype during the interruption. In summary, interrupting therapy prior to initiating salvage therapy in patients with advanced disease did not result in an improved virological response to enfuvirtide. The collective predictive activity of an enfuvirtide-containing regimen was important in predicting treatment response.