Definition of the time of loss of virological response is important when designing a viral load endpoint for trials of antiretroviral drugs. We assessed whether, in patients who achieved a viral load below 50 copies/ml, two consecutive values above 50 copies/ml was really indicative of loss of response. It was common for the viral load to return to below 50 copies/ml with no change in regimen, suggesting that a higher threshold for defining the loss of response is required.