The main goal of this study was to assess the role of mycophenolate mofetil (MMF) during interruption of highly active antiretroviral therapy (HAART). Seventeen patients with early-stage chronic HIV type 1 infection were treated with HAART for 12 months. They were then randomized (day 0) to receive MMF (HAART-MMF group, n = 9) or to continue the regimen (HAART group, n = 6) for 120 additional days. At day 120 in the HAART-MMF group, HAART was discontinued, and MMF administration was continued. The primary end point of the study was the number of individuals maintaining a plasma viral load (VL) set point of <200 copies/mL after at least 6 months off HAART. At day 120, all patients in both groups had undetectable plasma VLs. After 6 months off HAART, 5 of 9 patients in the HAART-MMF group versus 1 of 6 patients in the HAART group maintained a plasma VL of <200 copies/mL (P = 0.28). According to the ability of their plasma to inhibit cellular proliferation, patients were reclassified and divided into an inhibition group (n = 6) and a no inhibition group (n = 9). The doubling time of VL rebound was significantly higher in the inhibition group (mean +/- SE, 10.22 +/- 1.3) than in the no inhibition group (mean +/- SE, 4.6 +/- 1.6; P = 0.03). Moreover, 5 of 6 patients in the inhibition group maintained a plasma VL of <200 copies/mL versus 1 of 9 patients in the no inhibition group (P = 0.005) after 6 months off HAART. We found that combining MMF and HAART delayed VL rebound and improved control of viral replication without HAART but only when inhibition of lymphocyte proliferation was achieved.