OBJECTIVES: To compare two analytic approaches to assess the virological effect of HAART according to the intention-to-treat (ITT) principle. MATERIAL: Data from 2318 patients enrolled in 10 randomised clinical trials (RCTs) and from 3091 patients followed in an observation cohort (EuroSIDA) starting their first HAART regimen. METHODS: Two classifications of defining virological response 48 weeks after starting the therapy to be evaluated were compared: 1) only patients remaining on the therapy and having a plasma viral load (pVL) below a given cut-off level at week 48 were classified as responders (ITT/s=f); and 2) patients with a pVL below a given cut-off at week 48 whether they remained on initial assigned therapy or switched therapy were responders (ITT/s incl). In both analyses, patients with missing data at week 48 were classified as failures (i.e., non-responders). RESULTS: According to ITT/s=f, 22-70% of the patients starting a HAART regimen in a RCT experienced a virological response at week 48. Only two RCTs had complete follow-up data (n=424): between 29 and 62% achieved a virological response at week 48 in the six treatment arms evaluated in the studies according to ITT/s=f, and 41-72% according to ITT/s incl. Among those who discontinued the therapy to be evaluated in these two trials, 13-45% (cohort: 39-74%) subsequently experienced a virological response at week 48. The subsequent response rates were associated with the reason for discontinuation (toxicity versus confirmed virological failure: 63 vs 33%), varied largely across regimens and were not associated with the discontinuation rate. CONCLUSIONS: Discontinuation of follow-up at switch from the therapy to be evaluated remains common in antiretroviral treatment trials, but leads to an imprecise and incomplete assessment of the intrinsic effect of a given regimen. Complete follow-up of all patients should be encouraged strongly as this will allow for several complementary analytic approaches and a focus on optimal treatment strategies rather than specific regimens.