Objective: To determine the effectiveness and safety of adding delavirdine mesylate to a treatment regimen that included indinavir and nucleoside analog reverse transcriptase inhibitors in patients in whom combination therapy had failed. Design: Observational study. Setting: Private practice. Patients: HIV-1-positive patients with peripheral blood CD4+ lymphocyte counts < 300 x 10(6)/l in whom antiretroviral therapy had failed or whose condition was deteriorating. Of the 53 patients who were eligible for the study, 47 took part; for the majority of these patients indinavir combination therapy had been unsuccessful. the majority of the patients were male (98%), white (92%) and homo/bisexuals (96%). Interventions: Delavirdine added to current therapy (usually zidovudjne, indinavir and lamivudine); in approximately half of the patients zidovudine was replaced with stavudine. Main outcome measures: Plasma HIV-1 RNA levels; peripheral blood CD4+ and CD8+ lymphocyte counts. Safety of the therapy was assessed by monitoring side-effects. Results: Mean baseline CD4+ lymphocyte count was 127 x 10(6)/l and mean baseline HIV-1 plasma RNA was 5 log(10) copies/ml. Adding delavirdine to the therapeutic regimen produced a rapid and sustained decrease in the mean plasma HIV-1 RNA of 1.1 log(10) copies/ml over 6 months; 18-21% of patients showed decreases of 2-3 log(10) copies/ml. Viral burden in 33% of subjects declined below the assay's limit of detection (2.6 log(10) copies/ml) after 6 months. CD4+ lymphocyte counts increased by 66-90% in each group between 1 and 9 months (mean increase approximately 60 x 10(6)/l after 6 months). Adding delavirdine to current therapy was well tolerated. Side-effects reported were: skin rash, 28%; nausea, 9%; kidney stones, 9%; diarrhea, 6%; flank pain, 2%; proteinuria, 2%. Three patients reported serious medical events all of which resolved and none of which were attributed to delavirdine. Conclusions: Adding delavirdine to the combination regimen of patients in whom protease inhibitor therapy had failed often resulted in a rapid and remarkable decrease in viral load, sustained improvement in CD4+ lympocyte counts and viral load, and clinical improvement with minimal toxicity. (C) 1998 Lippincott-Raven Publishers.