Wilms tumor (WT) commonly occurs in infants and children. We evaluated clinical factors and the expression of multiple RNAs in WT samples in the TARGET database. Eight long non-coding RNAs (lncRNAs; AC079310.1, MYCNOS, LINC00271, AL445228.3, Z84485.1, AC091180.5, AP002518.2, and AC007879.3), two microRNAs (miRNAs; hsa-mir-152 andhsa-mir-181a), and nine messenger RNAs (mRNAs; TCTEX1D4, RNF133, VRK1, CCNE1, HEY1, C10orf71, SPRY1, SPAG11A, and MAGEB18) were screened from differentially expressed RNAs and used to construct predictive survival models. These models showed good prognostic ability and were highly correlated with tumor stage and histological classification. Additionally, survival-related ceRNA network was constructed using 35 RNAs (15 lncRNAs, eight miRNAs, and 12 mRNAs). KEGG pathway analysis suggested the “Wnt signaling pathway” and “Cellular senescence” as the main pathways. In conclusion, we established a multinomial predictive survival model and a survival-related ceRNA network, which provide new potential biomarkers that may improve the prognosis and treatment of WT patients.