Vincristine plus steroid pulses have long been a part of maintenance treatment in many protocols for childhood acute lymphoblastic leukaemia (ALL). A collaborative individual patient data meta-analysis of all randomised trials of the addition of vincristine plus prednisone/prednisolone (VP) pulses in childhood ALL was updated and extended to include trials comparing vincristine plus dexamethasone (VD) pulses to maintenance without pulses. VP pulses improved event-free survival (EFS) (70.1% vs. 62.0% at 5 years; odds ratio (OR) = 0.71; 95% confidence interval (CI) = 0.61-0.84; P = 0.00004); VD pulses did not have a significant effect (80.9% vs. 79.9% 5 year EFS; OR = 0.94; 95% CI = 0.80-1.11; P = 0.5). Heterogeneity between groups (VP or VD) was significant (P = 0.02). Neither treatment clearly affected overall survival. The difference between the VP and VD results is probably due to the greater early intensity of the backbone of the VD trial protocols and improved outcome seen in the VD trials, which were more recent. Pulses may still be useful in cases where less intensive early therapy is used and the balance between these treatments in terms of both effectiveness and toxicity needs to be considered.