Background: It is uncertain whether the same criteria for active surveillance can be applied universally across races. This population-based study was undertaken to quantify racial differences in long-term risk of prostate cancer-specific mortality (PCSM) among patients with low-risk prostate cancer (PCa) receiving conservative management. Methods: We used the Surveillance, Epidemiology, and End Results (SEER) database to identify patients who had low-risk PCa (T1-T2a & Gleason 6 & PSA ≤ 10 ng/mL & N0 & M0) diagnosed in 2004 – 2015 and did not receive radical prostatectomy or radiation therapy within one year of diagnosis. Kaplan-Meier analysis was used to calculate PCSM. The Clopper-Pearson method was used to calculate associated 95% confidence intervals. Hazard ratio of PCSM among those with a high PSA (PSA 4-10) compared to those with a low PSA (PSA < 4) was calculated using Cox proportional hazards models adjusted for covariates (including age, race, marital status, insurance status, U.S. region, year of diagnosis, and AJCC clinical tumor stage). Results: Among 33,740 patients with low-risk PCa, long-term PCSM varied with race and PSA levels at diagnosis. For instance, 10-year PCSM was 2.62% (95% CI: 1.15%-5.05%) among African Americans with PSA 4-10 and 0.98% (95% CI:0.16%-3.12%) among Caucasian patients with PSA < 4. There was no significant statistical interaction between race and PSA level on PCSM (p = 0.81). After adjusting for potential confounders, men with PSA 4-10 experienced 2-fold higher PCSM relative to those with PSA < 4 (HR = 1.96, p = 0.011) and African Americans men experienced a 43% higher PCSM compared to Caucasians (HR = 1.43, p = 0.03). Conclusions: Among men diagnosed with low-risk PCa, long-term PCSM varies by race and PSA at diagnosis. More refined risk stratification may improve PCa management among low-risk PCa patients.