Serum p24 antigen levels were examined in subjects from three clinical trials of zidovudine to determine whether the pattern of change in serum p24 antigen during the first 8-16 weeks of therapy was associated with human immunodeficiency virus type 1 (HIV-1) disease progression or death. Among 406 patients with AIDS and a first episode of Pneumocystis carinii pneumonia, 65% had measurable pretreatment concentrations of serum p24 antigen (≥10 pg/ml.). Changes during treatment were not associated with reduced mortality. In 637 mildly symptomatic patients, 24% had measurable concentrations, and changes were marginally associated with increased time until more advanced disease. Among 683 asymptomatic patients, 18% had measurable concentrations, and changes were not associated with increased time until progression. Despite the small number ofclinical events and the low rate of serum p24 antigen positivity in the latter two studies, pretreatment serum p24 antigen levels were predictive of clinical outcome; subsequent measurements appear to be of limited use in evaluating zidovudine therapy.