The present study was designed to screen key microRNA (miRNA)-target gene networks for ovarian cancer (OC) and to classify and construct a risk assessment system for OC based on the target genes. OC sample data of The Cancer Genome Atlas dataset and GSE26193, GSE30161, GSE63885 and GSE9891 datasets were retrospectively collected. Pearson correlation analysis and targeted analysis of miRNA and target gene were performed to screen key miRNA-target gene networks. Target genes associated with the prognosis of OC were screened from key miRNA-target gene networks for consensus clustering and least absolute shrinkage and selection operator-based regression machine learning analysis of OC samples. Twenty target genes of 2651 key miRNA-target gene pairs had significant prognostic correlation in each OC cohort, and OC was divided into three clusters. There were differences in prognostic outcome, biological pathways, immune cell abundance and susceptibility to immune checkpoint blockade (ICB) therapy and anti-tumor drugs among the three molecular clusters. S2 exhibited the least advantage in prognosis and immunotherapy response rate in the three molecular clusters, and the pathways regulating immunity, hypoxia, metabolism and promoting malignant progression of cancer, as well as infiltrating immune and stromal cell population abundance, were the highest in this cluster. An eight-target gene prognostic model was created, and the risk index obtained by using this model not only significantly distinguished the immune characteristics of the sample, but also predicted the response of the sample to ICB treatment, and helped to screen 36 potential anti-OC drugs. The present study provides a classification strategy for OC based on prognostic target genes in key miRNA-target gene networks, and creates a risk assessment system for predicting prognosis and response to ICB therapy in OC patients, providing molecular basis for prognosis and precise treatment of OC.