Ionizing radiation and a variety of genetic conditions are thought to explain 5-10% of childhood cancers. Infection with Epstein-Barr virus (EBV) in parts of Africa and human immunodeficiency virus (HIV) increase the risk of Burkitt's lymphoma and Kaposi's sarcoma, respectively. Other risk factors have not been conclusively identified. A review of the data on international variation in incidence, recent changes in incidence, and risk factors suggests that many childhood cancers are likely to have nongenetic causes. The pattern of international variation and associations with surrogates of infection suggest an infectious etiology for acute lymphoblastic leukemia, although no agent has been identified. The biologic plausibility is strong that maternal consumption of food containing DNA topoisomerase II inhibitors may increase the risk of acute myeloid leukemia, although the data are limited now. For brain tumors, cured meats, polyomaviruses, and farm exposures may have etiologic roles. Changes in the incidence and characteristics of children with hepatoblastoma as well as risk factor studies suggest a role for an exposure of very low birth weight babies. High birth weight, tea or coffee consumption, and certain paternal occupations have shown some consistency in their association with Wilms' tumor. For most of the other cancers, very few epidemiologic studies have been conducted, so it is not surprising that nongenetic risk factors have not been detected. The most important difference between the cancers for which there are good etiologic clues and those for which there are not may be the number of relevant studies.