Introduction: For people living with HIV, the first antiretroviral treatment (ART) regimen offers the best chance for a good virological response. Early identification of those unlikely to respond to first-line ART could enable timely intervention and increase chances of a good initial treatment response. In this study we assess the extent to which the HIV RNA viral load (VL) at 1 and 3 months is predictive of first-line treatment outcome at 6 months. Methods: All previously ART-naive individuals starting ART at two London centres since 2000 with baseline (−180 to 3 days) VL >500 c/mL had a VL measurement between 6 and 12 months after starting ART, and at least one at month 1 (4-60 days) or month 3 (61-120 days) were included. Lack of treatment response was defined as (i) VL >200 copies/mL at 6 months or (ii) VL >200 copies/mL at 6 months or simultaneous switch in drugs from at least two different drug classes before 6 months. The association with VL measurements at 1 and 3 months post-ART; change from pre-ART in these values; and CD4 count measurements at 1 and 3 months were assessed using logistic regression models. The relative fit of the models was compared using the Akaike information criterion (AIC). Results: A total of 198 out of 3258 individuals (6%) experienced lack of treatment response at 6 months (definition i), increasing to 511 (16%) for definition (ii). Those with a 1-month (day 4-60 window) VL of <1000, 1000-9999, 10,000-99,999 and >100,000 copies/ml had a 4%, 8%, 23% and 24% chance, respectively, of subsequently experiencing treatment non-response at 6 months (definition (i)). When considering the 3-month (day 61-120 window) VL, the chances of subsequently experiencing treatment non-response were, respectively, 3%, 25%, 67% and 75%. Results were similar for definition (ii). Conclusions: Whilst 3-month VL provides good discrimination between low and high risk of treatment failure, 1-month VL does not. Presence of a VL >10,000 copies/ml after 3 months of ART is a cutoff above which individuals are at a sufficiently higher risk of non-response that they may be considered for intervention.