Abstract OBJECTIVES: To assess the influence of patient characteristics, treatment precocity (how early) and duration of sustained virological response to highly active antiretroviral therapy (HAART) on HIV RNA levels after withdrawal of treatment started during primary infection and to compare HIV RNA levels after HAART withdrawal with levels reached at the same time point during the natural history of infection in the pre-HAART era. DESIGN: HIV RNA was analysed using linear mixed-effects models for 58 patients from the PRIMO cohort (1996-2003) treated during primary infection (with sustained virological responses until HAART interruption) and 116 untreated patients enrolled in the SEROCO cohort within 6 months following infection (1988-1995). Viral loads were estimated in PRIMO patients 36 months after infection (12 months after treatment interruption) and were estimated for the SEROCO patients 36 months after infection, after adjustment for gender and age. RESULTS: HIV RNA levels 12 months after HAART interruption were independently associated with levels at HAART initiation and with the CD4 cell count at HAART interruption, but not with the precocity of HAART or the duration of virological response to HAART. Thirty-six months after infection, mean HIV RNA levels were 3.95 log10 copies/ml 12 months after stopping HAART and 4.11 log10 copies/ml in never-treated patients. CONCLUSION: Viral load 12 months after withdrawal of transient effective HAART started during primary infection is similar to viral load at the same time after infection in never-treated patients, suggesting that early HAART initiation does not lower the virological set-point.