Hepatocellular carcinoma (HCC) is one of the malignancies with high morbidity and mortality in the world. The purpose of our study was to search for HCC related miRNA prognostic biomarkers to predict the risk degree and survival time of HCC patients and provide effective prognostic information for HCC patients. Four methods were used to identify differentially expressed miRNAs (DEMs) from The Cancer Genome Atlas(TCGA). Kaplan-Meier survival curve, univariate and multivariable Cox regression analysis were used to identify prognostic miRNAs of HCC from DEMs. Four prognostic miRNAs biomarkers (hsa-miR-132-3p, hsa-miR-139-5p, hsa-miR-3677-3p, hsa-miR-500a-3p) of HCC were identified at last, and combined into a risk score model. There was no experimental evidence that hsa-mir-3677-3p is related to HCC, and it was a newly discovered miRNA in this study. The evaluation results of various bioinformatics methods, including survival curve, ROC curve, chi-square test, et al.. All indicated that the risk score calculated by the model can effectively predict the risk degree of patients(P<0.000, hazard ratio=2.551, 95% confidence interval=1.751-3.717). 1-5 year survival rates of HCC patients in the low risk group had 20%-30% higher than in the high risk group. Through the clinical data analysis, it was found that the combined biomarkers have a better prognostic effect than other clinical indicators, and can also be used as an independent prognostic factor. Target genes of four miRNAs were predicted, including AGO2, FOXO1, ROCK2, RAP1B, CYLD, et al., and enriched in biological processes such as cell proliferation, migration, apoptosis and immune response.