We evaluated saquinavir, an orally active, selective inhibitor of HIV proteinase, in a randomised, double-blind, dose-ranging study in 49 zidovudine-naive HIV-positive patients with few or no symptoms and CD4 cell counts of 500 or less. The study was designed to assess the antiviral activity and tolerability of saquinavir. Patients were randomised to receive 25, 75, 200, or 600 mg of saquinavir three times daily for 16 weeks. No serious adverse events occurred. CD4 cell counts showed a trend indicative of a dose response in favour of the 600 mg dosage, the maximum increase being seen around week 4. In none of the 8 patients with positive plasma viraemia at baseline did cultures become negative after treatment; peripheral blood mononuclear cell and plasma-viral load by culture and DNA and RNA PCR all showed a trend towards reduction at higher doses of saquinovir. Saquinavir was well tolerated in this group of previously untreated patients with few or no symptoms; this study shows that an HIV-proteinase inhibitor is active in HIV-infected patients.