OBJECTIVES AND DESIGN: To investigate the clinical consequences of occasional and short (<or= 3 months) treatment interruptions in patients having initiated highly active antiretroviral therapy (HAART). Data from the prospective Swiss HIV Cohort Study were used. METHODS: Four different endpoints [death, Centers for Disease Control and Prevention (CDC) stages B and C, and CD4 cell count increase >or= 50 x 106/l] were studied in relation to the number of interruptions that occurred. In order to focus on short interruptions exclusively, observations of patients with a treatment interruption of > 3 months were censored. The CD4 cell count and viraemia were treated as time-dependent variables because of the importance of these factors when an interruption occurs. RESULTS: Between 1 January 1996 and 31 October 2000, 4720 Swiss HIV Cohort Study participants initiated HAART, which was interrupted at least once by 1299 participants. The main reasons for the interruptions were social factors. Interruptions did not increase significantly the risk of HIV-associated morbidity and mortality, except for a marginally increased risk for a CDC stage C event after the first interruption. The first interruption decreased significantly the likelihood of increasing the CD4 cell count. Subsequent interruptions had no further significant effect. High CD4 cell count and low viraemia, assessed as baseline and as longitudinal variables, were associated with a decreased risk of clinical progression. CONCLUSIONS: Occasional treatment interruptions of < 3 months neither worsen nor improve disease outcome on an average term (3-4 years). Our results suggest that interruptions might be non-risky, particularly when viraemia is low and CD4 cell count is high. These results require confirmation.