Two lines of attack have been used against HIV-1 disease: (1) antiretroviral therapy (notably zidovudine) directed against the HIV-1 virus; and (2) chemoprophylaxis against end-stage diseases of HIV-1 infection, most notably against Pneumocystis carinii pneumonia. Many studies and clinical trials have found that zidovudine and P. carinii prophylaxis each significantly reduce short term morbidity and mortality from HIV-1 disease. However, these drugs have costs, adverse effects and (particularly with zidovudine) cumulative toxicity that suggest that their long term benefits may not necessarily be as great as their short term prospects. Due to the pressing needs of the HIV-1 epidemic, social considerations and proven short term benefits, long term clinical trials of zidovudine and P. carinii prophylaxis employing untreated control groups are impossible. Thus, the long term benefits of these anti-AIDS therapies must be estimated from observational studies comparing those who choose to use those therapies and those who do not. Strong epidemiological biases and common statistical misanalyses of observational data occur, leading to seemingly contradictory results. On the basis of the aggregate of short term clinical data and long term observational studies, zidovudine therapy taken either before or after AIDS delays the date of death by 12 months. If zidovudine therapy is initiated before AIDS, it may delay the date of diagnosis of AIDS by 12 months. However, issues of cost effectiveness, detrimental effects and optimal time to begin zidovudine therapy remain unresolved. Limited observational study data suggest that P. carinii prophylaxis initiated before the onset of clinical AIDS will delay both the date of diagnosis of AIDS and the date of death by 9 months in North American men. The net benefits from P. carinii prophylaxis will be less in those who begin this treatment after a diagnosis of AIDS. Benefits from P. carinii prophylaxis will also probably be less in regions outside of North America, in particular underdeveloped countries, where P. carinii pneumonia is less common as an end-stage HIV-1-related illness.