We have examined the factors influencing prognosis in over 4000 children with acute lymphoblastic leukaemia (ALL) aged 1-14 who have been treated on consecutive MRC UKALL trials from 1972 to 1990. During this time the results of treatment have improved steadily but are consistently superior in girls when compared with boys; the 5-year event-free survival in girls improving from 51% to 71% and in boys from 31% to 57%. These results were independent of age and presenting leucocyte count. Boys not only had a testicular relapse rate of 10% but an excess of bone marrow relapse, particularly evident after 2 years from diagnosis. Other prognostic factors included organomegaly and the morphology of leukaemic blast cells; immunophenotype of the leukaemia, however, had no independent significance after allowance for age, sex and leucocyte count. The influence of sex on prognosis was reaffirmed when we examined various methods of identifying children at highest risk of treatment failure for whom alternative therapy such as bone marrow transplantation might be justified. In MRC UKALL X children had been deemed 'high risk' on the basis of leucocyte count alone, but with further follow-up it has become apparent that girls with an initial leucocyte count of > 100 x 10(9)/l have a similar prognosis to boys with a lower count. We therefore derived a risk score based on sex, age and count which has given better discrimination between standard risk (66% 5-year survival) and poor risk (39%) survival than other methods. This group of worse-risk children includes 16% of boys but only 3% of all girls. Gender remains an important prognostic factor in UKALL trials and there are very few girls who are at highest risk of treatment failure. The reasons for this remain unclear, but the pattern of relapses suggests that boys more often get inadequate systemic therapy. We postulate that the reasons for treatment failure may relate to sensitivity to continuing (maintenance) chemotherapy.