OBJECTIVES: To assess the safety of discontinuation of primary prophylaxis in HIV-infected patients on antiretroviral combination therapy at high risk of developing Pneumocystis carinii pneumonia. DESIGN: Prospective multicentre study. PATIENTS AND METHODS: The incidence of P. carinii pneumonia after discontinuation of primary prophylaxis was studied in 396 HIV-infected patients on antiretroviral combination therapy who experienced an increase in their CD4 cell count to at least 200 x 10(6)/l and 14% of total lymphocytes; the study population included 191 patients with a history of CD4 cell counts below 100 x 10(6)/l (245 person-years) and 144 patients with plasma HIV RNA above 200 copies/ml (215 person-years). RESULTS: There was one case of Pneumocystis pneumonia, an incidence of 0.18 per 100 person-years [95% confidence interval (CI), 0.005--1.0 per 100 person-years]. No case was diagnosed in groups with low nadir CD4 cell counts (95% CI, 0--1.2 per 100 person-years) or detectable plasma HIV RNA (95% CI, 0--1.4 per 100 person-years). CONCLUSIONS: Discontinuation of primary prophylaxis against Pneumocystis pneumonia is safe in patients who have responded with a sustained increase in their CD4 cell count to antiretroviral combination therapy, irrespective of the CD4 cell count nadir and the viral load at the time of stopping prophylaxis.