An understanding of the natural history of changes in prostate-specific antigen (PSA) may be valuable as a surrogate view of prostate dynamics, as a method to differentiate between benign and malignant growth, and as a means to assess the use of PSA as a tool for monitoring activity of chemoprevention agents. Although PSA appears to be useful as a noninvasive marker of prostatic growth, PSA changes should not be confused with a direct measure of tumor growth. Serum PSA levels are a function of tumor volume but are also influenced by the volume of benign epithelium, grade of carcinoma (if any), inflammation, androgen levels, growth factors, and the extracellular matrix. The biological functions of PSA in the prostate and in its secretions need to be more completely elucidated in order that PSA measurements may more accurately describe prostate dynamics. The expression of PSA is androgen-regulated. It is one of the most abundant prostate-derived proteins in the seminal fluid. Seminogelin, a major protein in seminal fluid, is cleaved by PSA, and this cleavage is important in the liquefaction of semen. Less is known about other PSA substrates. Current PSA studies indicate that cancer cases exhibit an early slow linear PSA phase followed by a rapid exponential phase, and that PSA levels begin to increase exponentially approximately 7-9 years before diagnosis. The establishment of age-specific PSA reference ranges (ASRR) and of PSA velocity (PSAV) rates provide elements of a baseline from which prediction models could measure malignant potential of a prostatic carcinoma. Moreover, recent discoveries of different molecular forms of PSA in serum may allow a much more accurate differentiation of benign and malignant growth as well as a more potent measure of the impact of chemoprevention agents. If PSA doubling time is approximately 2.4-3.0 years and accurately reflects tumor doubling time, and if the average man has less than 0.5 ml of latent prostatic tumor tissue and the average stage T2 cancer is approximately 4 ml when detected, then the available PSA data suggest that the 3 doublings necessary to change from 0.5-4.0 ml, would take 7-12 years for a typical small volume tumor to reach the size of most stage T2 tumors. The findings that histologic cancers appear at much younger ages than previously known is disturbing. It indicates that disease initiation may begin sooner than ever thought likely. "Normal" PSA levels for younger men (< 40 years of age) may need to be studied, and an emphasis upon premalignant lesions in this age group may be necessary. Younger men may represent the most appropriate population and premalignant lesions the most relevant clinical factor for prostate cancer chemoprevention studies and trials. The molecular composition and molecular changes of PSA derived from premalignant lesions have yet to be elucidated, but such investigations may lead to a more complete understanding of the possible progression or transformation of normal prostate cells to premalignancy and subsequently to carcinoma. High grade prostatic intraepithelial neoplasia (PIN) in and of itself does not account for elevated serum PSA levels, but subtle changes in the molecular dynamics of PSA may reveal the influence of androgens and the impact of chemopreventive agents.