Recent clinical trials have renewed interest in treating HIV-1 infection through the use of induction–maintenance regimens, a strategy commonly used for the treatment of TB and certain hematopoeitic malignancies. In these conditions, the common element underlying success has been the ability of induction regimens to reduce therapy-resistant pathogens to levels that can be controlled using a maintenance regimen. The success of recent clinical trials of induction–maintenance therapy for HIV-1 suggests that this concept could have a variety of applications, such as reducing toxicities, reducing treatment costs and improving the treatment of salvage patients. However, current induction–maintenance protocols have not fully capitalized on available quantitative data concerning pharmacokinetics and pharmacodynamics, viral replication dynamics, viral latency and the evolution of drug resistance. In this review, we reason that it should be possible to improve success rates of induction–maintenance and other innovative, therapeutic strategies using mathematical models that account for this information.