To assess the utility of quantitative competitive-polymerase chain reaction (QC-PCR) measurements of plasma human immunodeficiency virus type 1 (HIV-1) RNA and other viral load markers for assessment of antiretroviral therapy, we used archived cryopreserved specimens from a randomized controlled clinical trial of 135 patients (CD4+ T cell count < or = 500/mm3), comparing zidovudine (500 mg/day) versus the nonnucleoside reverse transcriptase inhibitor L-697, 661 (50, 300, or 1,000 mg daily). We evaluated treatment-associated changes in plasma viral load by standard and immune complex-dissociated (ICD) HIV-1 p24 antigen assays, and, in a representative subset of patients (n = 46), by QC-PCR determination of virion-associated HIV-1 RNA. At baseline, HIV-1 RNA was quantifiable by QC-PCR in all patients tested (100%), whereas standard and ICD HIV-1 p24 antigen tests were positive (> or = 30 pg/ml) in 42% and 56%, respectively. All viral load parameters showed significant decreases from baseline within 1 week of initiation of zidovudine, as measured by standard p24 antigen assay, ICD p24 assay, and QC-PCR. At 1 week, patients treated with either 300 or 1,000 mg/day of L-697,661 showed significant decreases from baseline in plasma standard and ICD p24 antigen and QC-PCR-determined HIV-1 RNA levels. Whereas viral load decreases seen with zidovudine were sustained for the duration of treatment, plasma viral markers often returned to pretreatment levels despite ongoing L-697,661 treatment, with evidence of the emergence of drug-resistant virus. Whereas standard p24, ICD p24, and viral RNA levels changed similarly in response to treatment, the superior sensitivity and available dynamic range of plasma viral RNA assays like QC-PCR analysis provide an advantage for clinical monitoring of plasma viral load, allowing tracking of treatment-related changes even in patients with earlier stage disease and lower levels of viral load.