BACKGROUND: It is unclear whether primary prophylaxis against Pneumocystis carinii pneumonia can be discontinued in patients infected with the human immunodeficiency virus (HIV) who are successfully treated with combination antiretroviral therapy. We prospectively studied the safety of stopping prophylaxis among patients in the Swiss HIV Cohort Study. METHODS: Patients were eligible for our study if their CD4 counts had increased to at least 200 cells per cubic millimeter and 14 percent of total lymphocytes while they were receiving combination antiretroviral therapy, with these levels sustained for at least 12 weeks. Prophylaxis was stopped at study entry, and patients were examined every three months thereafter. The development of P. carinii pneumonia was the primary end point, and the development of toxoplasmic encephalitis the secondary end point. RESULTS: Of the 262 patients included in our analysis, 121 (46.2 percent) were positive for IgG antibodies to Toxoplasma gondii at base line. The median CD4 count at study entry was 325 per cubic millimeter (range, 210 to 806); the median nadir CD4 count was 110 per cubic millimeter (range, 0 to 240). During a median follow-up of 11.3 months (range, 3.0 to 18.8), prophylaxis was resumed in nine patients, and two patients died. There were no cases of P. carinii pneumonia or toxoplasmic encephalitis. The one-sided upper 99 percent confidence limit for the incidence of P. carinii pneumonia was 1.9 cases per 100 patient-years (based on 238 patient-years of follow-up). The corresponding figure for toxoplasmic encephalitis was 4.2 per 100 patient-years (based on 110 patient-years of follow-up). CONCLUSIONS: Stopping primary prophylaxis against P. carinii pneumonia appears to be safe in HIV-infected patients who are receiving combination antiretroviral treatment and who have had a sustained increase in their CD4 counts to at least 200 cells per cubic millimeter and to at least 14 percent of total lymphocytes.