Background. Cervical cancer (CC) is a major malignancy affecting women worldwide, with limited treatment options for patients with advanced disease. The aim of this study was to identify novel prognostic biomarkers for CC. Methods. RNA-Seq data from four Gene Expression Omnibus datasets (GSE5787, GSE6791, GSE26511, and GSE63514) were used to identify differentially expressed genes (DEGs) between CC and normal cervical tissues. Functional and enrichment analyses of the DEGs were performed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database and the Database for Annotation, Visualization, and Integrated Discovery (DAVID). The Oncomine database, Cytoscape software, and Kaplan-Meier survival analyses were used for in-depth screening of hub DEGs. The Cox regression was then used to develop a prognostic signature, which was in turn used to create a nomogram. Results. A total of 207 DEGs were identified in the tissue samples, eight of which were prognostically significant in terms of overall survival (OS). Thereafter, a novel four-gene signature consisting of DSG2, MMP1, SPP1, and MCM2 was developed and validated using stepwise Cox analysis. The area under the receiver operating characteristic (ROC) curve (AUC) values were 0.785, 0.609, and 0.686 in the training, verification, and combination groups, respectively. The protein expression levels of the four genes were well validated by the western blotting. Moreover, the nomogram analysis showed that a combination of this four-gene signature plus lymph node metastasis (LNM) status effectively predicted the 1- and 3-year OS probabilities of CC patients with accuracies of 69.01% and 83.93%, respectively. Conclusions. We developed a four-gene signature that can accurately predict the prognosis in terms of OS, of CC patients, and could be a valuable tool for designing treatment strategies.