Highly active antiretroviral therapies (HAART) lower morbidity and mortality of HIV infection, but are unable to eradicate HIV and may cause side-effects. Planned interruptions of HAART are studied for three reasons: (1) to stimulate the anti-HIV immune response after viraemia has been suppressed by treatment; (2) to increase time off drug, to improve quality of life and diminish side effects and costs; and (3), among individuals whose virus has become resistant to treatment, to induce reversion of resistance to wild-type and therefore to improve the chances of success of subsequent salvage therapy. Regarding immune stimulation, the most promising results come from rare patients who started HAART during primary HIV infection. Up to 60% controlled viraemia below 5,000 copies/mL for up to 1 year without HAART. Among the majority of patients who start HAART later, during chronic HIV infection, preliminary results suggest that after repeated cycles on and off therapy, about 20% remain with a viraemia below 5,000 copies/mL after 12 weeks' treatment interruption. For the 80% who rebound to higher viraemia, additional immune-stimulatory manoeuvres are being considered such as administration of cytokines, or specific immune treatment by anti-HIV vaccines. When treatment is interrupted in patients who fail HAART and who have drug-resistant HIV, the drug-sensitive wild-type replaces the resistant quasispecies in 30-60% of patients. It remains to be seen whether this phenomenon increases the chances of successful salvage therapy. The risks and benefits of treatment interruption still have to be compared in large clinical trials, which are currently in the planning stage.