Objectives. To evaluate the effects of prednisone on prostate-specific antigen (PSA) in a cohort of patients with 'hormone-refractory' prostate cancer. Methods. Data were collected from 29 consecutive patients with hormone-refractory progressive prostate cancer who were treated with 10 mg of prednisone orally two times a day. Patients were included in this analysis only if other factors known to influence PSA levels (antiandrogen withdrawal, radiation, and/or other concomitant anticancer therapies) were definitively excluded as potentially confounding variables. Results. The mean and median PSA decline after initiating prednisone was 33% (95% confidence interval [CI] 20% to 46%) and 24% (range 0% to 99%), respectively. Ten patients (34%) had a PSA decline of more than 50% and 4 patients (14%) had PSA declines of more than 75%. The average and median time for progression-free survivals were 2.8 (95% CI 1.7 to 3.8) and 2.0 (range 0 to 11) months. Four (14%) patients had PSA declines lasting 6 months or more. Median survival was 12.8 months. Additional analyses indicated that a PSA decline of more than 50%, compared with less than 50%, was associated with a longer survival. Toxicities included steroid myopathy (n = 4), new-onset diabetes (n = 1), and dyspnea (n = 1). Conclusions. Prednisone (10 mg orally two times a day) can decrease PSA by more than 50% in approximately one third of patients with hormone- refractory progressive prostate cancer. On the basis of comparisons with other data sets, we hypothesize a dose-response relationship between glucocorticoid dose and PSA decline.