I have examined more than 800 values of serum prostate-specific antigen (PSA) in 119 American Veterans during the time before their diagnosis of prostate cancer. These values appear to follow an exponential model with respect to time. Specifically, the model comprises a sum of 2 exponential functions: one for an early, slowly rising component of PSA and a second for a later, faster rising component. The parameters of each component comprise an amplitude and a relative velocity. Whereas the relative velocity of the slow component is significantly associated with the volume of benign tissue, both the amplitude and relative velocity of the fast component are significantly associated with the volume of tumor. The results suggest that at the time of diagnosis of prostate cancer the level and velocity of PSA reflect the combination of slow and fast components. Thus, this model provides insight into how benign and malignant tissues in the prostate determine the dynamics of PSA.