Cervical cancer is the second cause of cancer death in females in their 20s and 30s, but there were limited studies about its prognosis. This study aims to identify miRNA related to prognosis and study their functions. TCGA data of patients with cervical cancer were used to build univariate Cox's model with single clinical parameter or miRNA expression level. Multivariate Cox's model was built using both clinical information and miRNA expression levels. At last, STRING was used to enrich gene ontology or pathway for validated targets of significant miRNAs, and visualize the interactions among them. Using univariate Cox's model with clinical parameters, we found that two clinical parameters, tobacco use and clinical stage, and seven miRNAs were highly correlated with the survival status. Only using the expression level of miRNA signatures, the model could separate patients into high-risk and low-risk groups successfully. An optimal feature-selected model was proposed based on two clinical parameters and seven miRNAs. Functional analysis of these seven miRNAs showed they were associated to various pathways related to cancer, including MAPK, VEGF and P53 pathways. These results helped the research of identifying targets for targeted therapy which could potentially allow tailoring of treatment for cervical cancer patients.