To evaluate the longitudinal change in prostate-specific antigen (PSA) and the influence of initial PSA on the PSA change. We retrospectively analysed health examination data collected at Beijing Hospital from March 2007 to November 2011. Men with an initial PSA levels less than 4 ng ml(-1) and an annual PSA test for 5 years were enrolled into the study. The men were separated into four groups by the initial PSA level (0-0.99, 1-1.99, 2-2.99 and 3-3.99 ng ml(-1)), and the difference in PSA change among the four groups was analysed. A total of 1330 men were enrolled into the study. The mean age, initial PSA and PSA velocity (PSAV) were 58.17 ± 14.63 (range 24-91) years, 1.18 ± 0.79 (range 0-4) ng ml(-1) and 0.04 ± 0.25 (range -1.34-2.02) ng ml(-1) year(-1). Pearson's correlation analysis showed no correlation between initial PSA and PSAV (r=-0.036, P=0.189). The PSAV of the 0-0.99, 1-1.99, 2-2.99 and 3-3.99 ng ml(-1) initial PSA groups was 0.03 ± 0.11, 0.07 ± 0.32, 0.03 ± 0.34 and -0.01 ± 0.43 ng ml(-1) year(-1), respectively (P=0.06). As the initial PSA increased, the percentage of having a PSAV over 0.75 ng ml(-1) year(-1) and a negative PSAV both significantly increased. Males with a baseline PSA of 0-0.99, 1-1.99, 2-2.99 and 3-3.99 ng ml(-1) had a 1.88%, 6.16%, 16.30% and 57.81% chance, respectively, that their PSA would increase above 4.0 ng ml(-1) over the following 4 years (P<0.0001). The PSAV has no correlation with the initial PSA level. However, as the initial PSA increases, the chance that males will have an abnormal PSA or PSAV in the future increases.