To the investigator and clinician, prostate-specific antigen (PSA) level is a seemingly perfect outcome measure because it is easily assessable, quantitative, reproducible, and inexpensive. Whether post-therapy decline in PSA reflects true clinical benefit, and whether post-therapy declines can be used as an intermediate endpoint for accelerated drug approval is still open to question. At present, no drug has been approved strictly on the basis of a post-treatment decline in PSA, as it is unproven that such PSA changes are surrogates for true clinical benefits. Post-therapy PSA changes have been associated with improved survival in patients with castrate metastatic disease. The role of PSA changes as potential surrogates of clinical benefit have only been explored to a limited degree because to date, only two prospective randomized trials showing a survival benefit have been reported. Such trials are necessary, but not a sufficient pre-requisite to explore the potential role of any outcome measure as an intermediate endpoint. The clear demonstration that a post-therapy PSA change can account for all of the treatment effects seen is not yet available. A cytotoxic drug that does not produce any PSA decline is unlikely to be effective, but the converse is not always true because not all PSA rises represent a treatment failure. It is important to recognize that there are a range of clinical benefits to patients that can improve the quality and possibly the duration of survival, independent of PSA.