The study of minimal residual disease (MRD) as a 'surrogate' marker of molecular response to treatment has drawn great interest because of the potential of tailoring treatment and the possibility of gaining insight into the nature of a cure. Polymerase chain reaction-based (PCR-based) detection of MRD by immunoglobulin (Ig) and T-cell receptor (TCR) gene rearrangements can be applied in more than 90-95% of cases of childhood acute lymphoblastic leukaemia (ALL). Accordingly, several retrospective studies of MRD in childhood ALL have used one of the different PCR approaches for the detection of antigen-receptor gene rearrangements. The promising results on the predictivity of MRD evaluation at the end of induction treatment has raised the need of a new definition of remission. Until now, most PCR-based MRD studies have used semiquantitative methods for the detection of Ig and TCR gene rearrangements. The introduction of real-time quantitative PCR (RQ-PCR) has resulted in the improvement of sensitivity and specificity and has given better quality control of the MRD data. There is an urgent need to incorporate MRD data in clinical studies, properly designed to address treatment questions. In this context several ongoing co-operative study groups have adopted an MRD-based risk group classification to explore whether a better tailored treatment would result in further improvement in cure rates for children with ALL.