OBJECTIVE: To assess the safety, efficacy and plasma drug levels of the combination of ritonavir plus saquinavir for the treatment of advanced HIV infection. DESIGN: Multicentre pilot study. PATIENTS: Eighteen protease inhibitor-naive patients, with intolerance or contraindication to reverse transcriptase inhibitors, a median CD4 cell count of 12 x 10(6)/l (range, 1-50 x 10(6)/l), and a median HIV viraemia of 5.25 log10 copies/ml (range, 4.00-6.13 log10 copies/ml). METHODS: Patients received 600 mg twice daily of both ritonavir and saquinavir. Viraemia was measured at baseline and at weeks 5, 9 and 13. Response was defined as a drop of viraemia of more than 1 log10 at week 5. Plasma drug levels were determined after at least 3 weeks of combined treatment: samples were collected before and 1, 2, and 4 h after the morning ingestion of both drugs. The protease gene was sequenced at baseline and under treatment. RESULTS: Among the 16 patients evaluable at week 5, 11 were responders, and among these patients, six remained responders at week 13 (two with undetectable viraemia). Study discontinuations were due to side-effects (n = 4), patient choice (n = 3), protocol violation (n = 1) and death (n = 1). Responders had higher drug levels than non-responders (P < 0.01 for saquinavir, P = 0.04 for ritonavir). In two non-responders, development of multiple new mutations at positions 10, 20, 48, 82, 84 and 90 was observed after 5-13 weeks. CONCLUSION: The response to ritonavir plus saquinavir in advanced HIV infection is unpredictable. A minority of patients respond with disappearance of HIV viraemia. In other patients, rapid cumulative emergence of protease mutations conferring resistance to treatment cannot always be prevented by good compliance and relatively high plasma drug levels.