Understanding the extent to which childhood leukaemia and non-Hodgkin lymphomas are heritable is important to the survivors of these diseases, their families and clinicians who provide genetic counselling. Such understanding is also relevant to the possibility raised by Gardner et al. (1990, Br. Med. J., 300, 423-429) that paternal preconception irradiation may be an aetiological factor in these diseases. No malignant neoplasm was diagnosed among 382 offspring of survivors of childhood leukaemia and non-Hodgkin lymphoma followed up for a median period of 5.8 years, the largest available cohort of such offspring. These data indicate that it is unlikely that the risk of a malignant neoplasm occurring in the offspring exceeds eight times that expected in the general population. Similarly, the risk of leukaemia and non-Hodgkin lymphoma among offspring is unlikely to exceed 21 times that expected. The proportion of survivors of childhood leukaemia and non-Hodgkin lymphoma with the heritable form of these diseases is unlikely to exceed 5%, assuming an autosomal dominant pattern of transmission, with penetrance of at least 70% and that all heritable cases develop by age 15 years. The best (i.e. at present most likely) estimates of these risks are of course much lower. There was no evidence of an excess of congenital abnormalities among the offspring and the sex ratio was similar to that expected from the general population.