The impact of various types of intensification therapy was examined in a cohort of 3617 children aged 1-14 years with acute lymphoblastic leukaemia (ALL) enrolled in the Medical Research Council (MRC) UKALL X (1985-90) and UKALL XI (1990-97) trials. UKALL XI was modified in 1992 to incorporate the "best arm" of UKALL X with two 5-d intensification blocks at 5 and 20 weeks, and an additional randomization in respect of a third intensification at 35 weeks but omission of two consecutive injections of daunorubicin during induction. All children were eligible for randomization irrespective of risk group. The impact of the various types of intensification therapy was examined in a stratified analysis. At a median follow up of 102 months, both trials had an identical event-free survival of 61% (95% CI 58-63%) at 8 years. Survival at 8 years in UKALL XI was significantly better in than in UKALL X, 81% (79-83%) compared with 74% (72-76%) (P = < 0.001), owing to improved management of relapse. There was a highly significant trend in reduction of the number of relapses and deaths with increased intensity of therapy both for children with initial leucocyte count < 50 x 10(9)/l (P = < 0.001) and > or = 50 x 10(9)/l (P = 0.002). Introduction of a third late intensification block compensated for omission of anthracyclines during induction but produced little additional benefit. These results show, in a large cohort of patients, that minor modifications of therapy may influence relapse rate and obviate the benefit of previous randomized trials. The failure to adapt treatment for higher risk children contributed to these disappointing results.