AIM: To explore the key genes and molecular markers involved in the retinoblastoma development through bioinformatics.METHODS: The mRNA microarray datasets from the Gene Expression Omnibus(GEO)database were obtained, and the differentially expressed gene(DEG)between retinoblastoma cell lines and normal retinal pigment epithelial(RPE)cell lines were analyzed through gene ontology(GO)and KEGG enrichment analysis. To screen key genes, establish protein-protein interaction(PPI)network, and use receiver operating characteristic(ROC)curve to assess clinical diagnostic efficacy. The RNA expressions of key genes in retinoblastoma cell lines and normal RPE cell lines were compared by qRT-PCR.RESULTS: A total of 121 DEGs were obtained from the retinoblastoma dataset of GSE97508 and GSE110811. KEGG pathway analysis showed that DEG were enriched in phototransduction, cell cycle, and p53 signaling pathways. A total of 9 key genes, including MCM6, DTL, UBE2T, TOP2A, NUSAP1, CENPK, RRM2, RLBP1, and RHO, were obtained from the intersection of PPI network analysis and the top 30 DEG from each dataset. The differentially expressed 9 key genes were verified in GSE24673. ROC analysis showed that the area under the curve(AUC)for UBE2T, RRM2, and RHO was ≥80%, and there was a statistical significance(P>0.05). The mRNA level of UBE2T and RRM2 in retinoblastoma was significantly higher than APRE-19 cell line, while the mRNA level of RHO was significantly lower than that of ARPE-19 cell line.CONCLUSION: UBE2T, RRM2, and RHO may be served as potential molecular markers and potential therapeutic targets for retinoblastoma.