Abstract Objectives. To investigate the relationship between age and total prostate-specific antigen (tPSA), free PSA (fPSA), and percent free PSA (%fPSA) in men 48 to 79 years old without clinical evidence of prostate cancer. We determined age-specific ranges for each parameter and compared the results with previously published studies in similar populations. Methods. Nine hundred eighty-three men (96% white) from the random-sample community-based Massachusetts Male Aging Study were analyzed. Men with PSA 4.1 ng/mL or greater were referred for biopsy and those with positive biopsies or with medical record, cancer registry, or self-reported evidence of prostate cancer were excluded. Results. The median tPSA increased 38.6% per decade (95% confidence interval 28.7% to 49.3%). Because of the greater variability at older ages, the 95th percentile increased faster than the median, leading to the following age-specific upper limits of normal: 2.84 for 50 to 59 years, 5.87 for 60 to 69 years, and 9.03 for 70 to 79 years. The pattern of association between fPSA and age was similar to tPSA. The 50th and 5th percentiles of %fPSA were 25.3% and 13.2%, respectively, regardless of age. Conclusions. Establishing age-specific screening cutoffs based on the age-specific upper limits of normal will ensure low false-positive biopsy rates but may also lead to low true positive rates (ie, low sensitivity) in older age groups. Both sensitivity and specificity should be considered when counseling patients. The independence of %fPSA with age confirms others’ findings.