The benefits of achieving a long term event free survival of 60-70% by using increasingly intense treatment regimens must be weighed against the increased risk of treatment toxicity. From 1985 to 1990, 1612 children with childhood acute lymphoblastic leukaemia (ALL) in the UK were treated on MRC UKALL X with intensive induction therapy, central nervous system directed therapy (cranial irradiation and intrathecal methotrexate), and continuing treatment for two years. There was a randomisation to receive blocks of additional intensification treatment at five weeks, 20 weeks, not at all, or both. The five year disease free survival was 71% for children randomised to two blocks of intensification, a 14% improvement on children randomised to no intensification treatment. Treatment related mortality in this national multicentre study has been analysed for induction and first remission (including those after intensification treatment). There were 38 induction deaths, 2.3% and 53 deaths in first remission, 3.3% (including those from a second malignancy). Thirty one (84%) of the induction deaths followed an infection: bacterial in 22 and fungal in nine. Thirty seven infective remission deaths occurred: bacterial in 11, viral in 16, fungal in seven, and three caused by Pneumocystis carinii pneumonia. Ten of these deaths followed a block of intensification treatment. The majority of noninfective remission deaths followed the development of a second tumour. Risk analysis for an induction death showed girls and children with Down's syndrome to be at greater risk. For deaths in first remission analysis showed an increased risk for bone marrow transplant (BMT) patients and children with Down's syndrome. There was no effect of age and leucocyte count for either group. Most significantly when BMT patients were excluded from the analysis, intensification treatment did not increase the risk of remission death.