Wilms tumor can be explained only partially by the "two hit" model that was originally developed for retinoblastoma. Heterogeneity of two kinds operates. The first is that four other primary tumors are regularly observed in children, and the second is that Wilms tumor itself appears to represent more than one genetic entity. All five of these primary renal tumors arise from primary or secondary mesenchyme, renal blastema, or renal epithelium. Mesoblastic nephroma, and possibly clear cell sarcoma, may have some genetic affinity with Wilms tumor, but rhabdoid tumor of the kidney and renal carcinoma do not. At least three different genes seem to be important in the origin of Wilms tumor. One, WT1, whose mutations may be associated with aniridia, may follow the "two hit" model in that there are cases in which both copies of the gene are defective or lost, as expected for a tumor suppressor gene. A second gene, which is associated with Beckwith-Wiedemann Syndrome (BWS) and which has not been cloned, appears to be imprinted in females, and may have an oncogene function. It is evidently activated by gain of a paternal allele or by loss of the inactive, but possibly trans-sensing, maternal allele. Activation of the insulin-like growth factor II gene may be a final common pathway for mutation in both WT1 and BWS. A third gene is unlinked to either of the other two, but its location and function are unknown. It shares with WT1 specificity for Wilms tumor, which is not true of the BWS gene.