Post-therapy changes in prostate specific antigen (PSA) have been proposed as a surrogate end point for response in clinical trials of patients with hormone refractory prostatic cancer. While the specific criteria for response are evolving, there are concerns that changes in serum levels of PSA may be the result of changes in PSA expression or secretion independent of effects on cell growth and proliferation. We report 3 representative cases receiving complete androgen blockade with either gonadotropin-releasing hormone or orchiectomy plus the antiandrogen flutamide, which demonstrated sustained declines in serum PSA levels after discontinuation of the antiandrogen. Similar results have been observed in 6 additional patients. Whether this represents an agonist effect of the antiandrogen is unknown. The results suggest that a trial of flutamide withdrawal is justified in an asymptomatic patient with an increasing PSA before treatment with more toxic therapies. It also shows the importance of documenting sequential elevations in the marker before enrolling a patient in a clinical trial, and that changes in PSA must be interpreted cautiously in clinical trials when an antiandrogen is discontinued.