The possibility that human growth hormone (GH) replacement therapy might either increase the risk of cancer recurrence in a child who has previously been treated for a brain tumour or leukaemia, or induce de novo cancer, has worried paediatricians for a number of years. Concern arises from animal experiments, the association of acromegaly with malignancy, and the Japanese experience of a cluster of de novo leukaemia cases in children treated with GH. It is reassuring that so far the results from single centre studies and from the pharmaceutical industry surveillance programmes have shown no evidence of an increased risk of malignancy, recurrent or de novo. The confidence intervals, however, are wide and the scientific nature of these studies is flawed as there has never been a prospective randomized study of GH replacement in children with radiation-induced GH deficiency. For clinical reasons, such a study is unlikely to be performed and therefore surveillance must be maintained at a very high level.