Background: Primary liver cancer is the fifth most common malignancy and the second most frequent cause of cancer-related death worldwide. Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, followed by Intrahepatic cholangiocarcinoma (ICC), and the incidence of liver cancer is rapidly rising in the United States. Race and ethnicity affect the outcomes in many solid cancers; therefore, we aimed to study the effect of race and ethnicity on overall survival (OS) in primary liver cancer. Methods: We selected adult patients who had primary liver cancer (HCC and ICC) diagnosed between 2004 and 2016 using the NCDB database. We stratified the patients into two groups as white and black. We performed the Kaplan-Meier method for median overall survival (OS), and the Cox regression method for multivariate analyses after adjusting for clinically significant confounding variables. Also, we used 1:1 nearest neighbor propensity score matching to reduce selection bias. Results: A total of 122,049 patients, 83.1% was white, 16.9% was black. Black patients were more likely to be young (p<.001), to live in the metropolitan area (p<.001) and South Atlantic states (p<.001), to be treated academic/research facility (p<.001), to be uninsured (p<.001). In contrast, white patients were most frequently to be highly educated (p<.001), to have higher income (p<.001), to travel longer distances (≥50 miles) for treatment (p<.001). Also, black patients had a higher rate of HCC (p<.001), advanced stage (p<.001), high metastasis rate (p<.001), whereas white patients had a higher rate of ICC (p<.001), higher comorbidity (p<.001), high treatment rate (surgery, chemotherapy, and radiotherapy) (p<.001). The gender percentage was similar in both groups. In unmatched groups, white patients had significantly better median OS compared to black (9.9 vs. 8.4 months, p<.001), but this difference did not continue after adjusted for age, sex, facility type, insurance status, income, education, rurality, travel distance, comorbidity, metastatic rate, treatment receipt (surgery, chemotherapy, radiotherapy), primary site (HR=1.01 95% CI: 0.99-1.03, p=.18). In ethnic groups, Hispanics had significantly improved OS compared to non-Hispanic white (NHW) and non-Hispanic black patients (NHB) (median OS 12 vs. 9.7 vs. 8.5 months, respectively, p<.001), and this significance was persisted with multivariate analysis (HR=0.80 95% CI: 0.78-0.82, p<.001). After propensity score matching, white and black patients were similar OS (p=NS) but Hispanics were still significantly improved OS versus NHW and NHB patients (HR=0.76 95% CI: 0.71-0.80, p<.001). Conclusions: Black patients had a higher stage, percentage of metastasis and lower treatment rate, but they had a similar OS compared to white patients. Hispanics had significantly improved OS than others, even after adjusted for clinically important confounding variables.