One of the most difficult problems in HIV care today is the management of individuals infected with multidrug-resistant viruses. Well-controlled, carefully designed clinical trials have recently resulted in the approval of several new antiviral agents for the treatment of drug-resistant HIV. The design of these trials has come at the cost of the predictable emergence of drug-resistant viruses among individuals randomly assigned to receive a suboptimal revised treatment regimen. We propose here a different approach to the evaluation of drug efficacy in individuals harbouring multidrug-resistant HIV.