PSA has been a valuable tool in enhancing our understanding of the prevalence and virulence of prostate cancer. PSA also has contributed to the understanding of important phenomena related to the androgen regulation of the cancer; however, it has not been useful in detecting some forms of androgen-independent (neuroendocrine) progression and is of limited prognostic value in androgen-independent prostate cancer. PSA also has been valuable in the accelerated development of therapies for prostate cancer; however, it must be used cautiously for this purpose, because it may not reflect the most relevant clone. In addition, some agents may directly affect PSA release independent of their antitumor activity. Most importantly, before PSA is adopted as a surrogate end point in clinical trials in prostate cancer, it must be prospectively validated. Future studies must focus on the development of prospective serologic tumor markers that can predict virulence of disease and to reflect androgen-independent progression.