Whether zidovudine (3'-azido-3'-deoxythymidine, AZT) should be offered to symptomless individuals infected with human immunodeficiency virus type-1 (HIV-1), in the hope of delaying or even preventing progression to AIDS, has been much debated. The discussion has focused on the efficacy of the drug in delaying progression to disease, the severity of its side-effects, and the likelihood of its prolonged and widespread use resulting in zidovudine-resistant strains of the virus. Little attention has been given to the degree to which treatment reduces the infectiousness of symptomless patients, and to the concomitant implications for the overall transmission rate of HIV-1 in the community. Here we use simple mathematical models to show that community treatment with antiviral drugs or immunotherapies that lengthen the incubation period of AIDS without significantly reducing the infectiousness of treated individuals, can increase the rate at which HIV-1 infection spreads (which is fairly obvious) and can even, under certain circumstances, increase the AIDS-related death rate in the community (which is less obvious).