OBJECTIVE To assess the short-term and long-term effect of a combination of saquinavir, ritonavir and stavudine in moderately to severely immunosuppressed protease inhibitor-naive patients. METHODS Prospective open-label multicentre study. METHODS A total of 64 protease inhibitor-naive and stavudine-naive HIV-infected patients with a CD4 count of < 250 cells/microL and > 10 000 HIV-1 RNA copies/mL received saquinavir hard-gelatin capsules, ritonavir and stavudine. Full (drop in viraemia of > 2 log10 and/or < 500 copies/mL) and partial responders (drop to between 500 and 5000 viraemia copies/mL) at week 9 (end of phase I) entered the second phase (additional 12-month period). RESULTS Fifty-six patients completed phase I, 45 (70%) full responders and nine (14%) partial responders by intent-to-treat analysis. Thirty-nine patients completed phase II, 33 (52%) full responders and two (3%) partial responders. Six patients had < 50 HIV-1 RNA copies/mL at week 9, and 20 (31%) patients at month 12 of phase II. Mean CD4 cell counts increased significantly in the 56 patients from 89 to 184 cells/microL after 9 weeks and from 100 to 292 cells/microL in the 39 patients treated for another 12 months. Higher baseline viraemia and lower baseline CD4 cell counts were not associated with an unfavourable virological response at week 9 and month 12 of phase II. HIV DNA in peripheral blood monocytes decreased substantially (- 1.5 log10) but was detectable in all except one patient at the end of phase II. CONCLUSIONS In protease- and stavudine-naive HIV-infected patients with moderate to severe immunosuppression, saquinavir in combination with ritonavir and stavudine caused a substantial long-term decrease in plasma viral load in approximately half the participants and a substantial increase in CD4 cell counts.