Your News Feature “Staying the course” (Nature 442, 617–619; 2006) offered a welcome dose of optimism in the debate about mass deployment of antiretroviral therapies in Africa, by arguing that this had not yet created an epidemic of drug-resistant HIV. But although everyone in the field has been delighted by initial studies suggesting high adherence, we must not be complacent.

The News Feature cited studies showing that 70–80% of patients on antiretroviral drugs had undetectable viral loads; the converse is that 20–30% have detectable viral loads. The spread of resistance is determined not only by compliance — which, as you point out, has been good so far in most settings — but also by factors more specific to Africa, such as interactions between HIV and TB drugs, or by limitations of drug-supply infrastructure. The predominant use of clinical monitoring in African programmes may allow mutations to accumulate (DART Virology Group AIDS 20, 1391–1399; 2006), potentially making resistant strains more transmissible. If virological failure accompanies the roll-out of these drugs, it could rapidly increase with mass deployment: see how fast this might happen at Scientists must pool data from roll-out cohorts (functioning as sentinel sites), not only to assess resistance but also to gather data on adherence, pharmacokinetics, drug interactions and other factors that determine a drug regimen's useful lifespan.

The consequences of widespread use of nevirapine to prevent mother-to-child HIV transmission are not yet clear, and a public health fall-back plan is urgently needed in case large-scale resistance emerges. Well-funded education programmes are vital to help people minimize their risk of infection, encourage them to undergo the counselling and testing they need to gain access to antiretrovirals, and overcome the stigma of AIDS. The success of initial antiretroviral deployment is encouraging, but we should not underestimate the challenges that remain.