OBJECTIVE To assess the HIV-1-RNA response and drug concentrations in cerebrospinal fluid (CSF) and serum during treatment with saquinavir (SQV)/ritonavir (RTV) or SQV/RTV plus stavudine (d4T) in HIV-1 -infected patients. METHODS A multicentre, open-label, randomized controlled trial. METHODS A total of 208 protease inhibitor (PI) and d4T-naive, HIV-1-infected patients were treated with RTV 400 mg twice daily and SQV 400 mg twice daily with or without d4T 40 mg twice daily. Intensification with reverse transcriptase inhibitors was allowed if serum HIV RNA remained above 400 copies/ml after 12 weeks. In 27 volunteers, CSF and serum HIV RNA were measured at baseline, weeks 12 and 48, using the Roche Amplicor and the ultrasensitive assay. In 22 patients, serum and CSF drug concentrations were determined at week 12. RESULTS The median baseline serum and CSF HIV-RNA concentrations were 4.81 and 3.21 log10 copies/ml, respectively. A difference in the proportion of patients with a CSF HIV-RNA level below the limit of quantification (< LLQ) after 12 weeks was found: four out of 14 (RTV/SQV) versus 12 out of 13 (RTV/SQV/d4T) (P = 0.001). The same results were found using the ultrasensitive assay. Patients with a baseline HIV-RNA level < LLQ in CSF remained < LLQ, regardless of the treatment regimen. Treatment with RTV/SQV alone was the only independent predictor of a CSF HIV-RNA level > LLQ at week 12 in logistic regression analysis (P = 0.005). CSF RTV and SQV concentrations were < LLQ in most patients. CONCLUSIONS RTV/SQV alone cannot suppress detectable CSF HIV-1-RNA levels to < LLQ after 12 weeks of treatment in the majority of patients. CSF drug concentrations of RTV and SQV < LLQ may explain the suboptimal antiretroviral effect in the CSF.