"Race-specific" prostate-specific antigen (PSA) needs evaluation in men at high risk for prostate cancer for optimizing early detection. Baseline PSA and longitudinal prediction for prostate cancer were examined by self-reported race and genetic West African (WA) ancestry in the Prostate Cancer Risk Assessment Program, a prospective high-risk cohort. Eligibility criteria were age 35 to 69 years, family history of prostate cancer, African American race, or BRCA1/2 mutations. Biopsies were done at low PSA values (<4.0 ng/mL). WA ancestry was discerned by genotyping 100 ancestry informative markers. Cox proportional hazards models evaluated baseline PSA, self-reported race, and genetic WA ancestry. Cox models were used for 3-year predictions for prostate cancer. Six hundred forty-six men (63% African American) were analyzed. Individual WA ancestry estimates varied widely among self-reported African American men. Race-specific differences in baseline PSA were not found by self-reported race or genetic WA ancestry. Among men with > or =1 follow-up visit (405 total, 54% African American), 3-year prediction for prostate cancer with a PSA of 1.5 to 4.0 ng/mL was higher in African American men with age in the model (P = 0.025) compared with European American men. Hazard ratios of PSA for prostate cancer were also higher by self-reported race (1.59 for African American versus 1.32 for European American, P = 0.04). There was a trend for increasing prediction for prostate cancer with increasing genetic WA ancestry. "Race-specific" PSA may need to be redefined as higher prediction for prostate cancer at any given PSA in African American men. Large-scale studies are needed to confirm if genetic WA ancestry explains these findings to make progress in personalizing prostate cancer early detection.