OBJECTIVE: To evaluate the efficacy of early initiation of highly active antiretroviral therapy (HAART), we compared the clinical course of two nested, matched cohorts within the Swiss HIV Cohort Study. METHODS: We selected all asymptomatic patients who started HAART between 1 January 1996 and 31 December 1999 with a CD4 cell count > 350 x 10(6)/l. We then matched them with asymptomatic participants who were seen at around the same time and who remained untreated during the following 12 months. This control group was further matched for age, sex, CD4 cell count, viral load, and HIV risk category, generating 283 pairs of treated versus untreated patients. RESULTS: During observation of median 3.19 versus 2.66 years, CDC stage B/C occurred in 6.4% versus 21.2%, AIDS in 1.8% versus 5.3%, death in 2.1% versus 6.4%, and AIDS or death of 'natural' causes in 2.8% versus 6.7% of the treated versus untreated patients. In multivariable Cox regression analysis, treatment reduced the risk of clinical progression by a factor of four- to five fold. During follow-up, the treated group had significantly higher CD4 counts and lower HIV-1 RNA levels. Intolerance/adverse events led to change or stop of at least one drug in 35% of treated patients. The entire regimen was interrupted at least once by 41% of patients, and 24% had no treatment anymore at the end of follow-up. CONCLUSIONS: The initiation of HAART in asymptomatic patients with CD4 cell count > 350 x 10(6)/l significantly delayed clinical progression. However, the risk of severe clinical events with deferred therapy was low and must be counter balanced against the burden and toxicity of HAART.