The follow up of minimal residual disease (MRD) in childhood B-precursor ALL by polymerase chain reaction (PCR) may be of help for further stratification of treatment protocols, to improve outcome. However, the clinical relevance of this approach has yet to be defined. We report the retrospective follow-up of MRD in bone marrow (BM) samples from 50 childhood B-precursor ALL patients by IgH/TCR delta PCR. Twenty-two patients remained in continuous complete remission (median follow-up 61 months), and 28 experienced relapse (median follow-up 75 months). Initial regression of MRD on therapy correlated with outcome. At the end of induction therapy 2/18 (11.1%) patients from the CCR group were PCR positive vs 10/16 (62.5%) from the 'relapse' group (P = 0.005). The presence of PCR detectable MRD predicted event-free survival independent of standard clinical and cytogenetical parameters. Also subsequent to first BM relapse, a correlation between MRD regression and outcome was observed. Six of eight patients who became PCR negative in the time period between relapse and bone marrow transplantation are in CCR, whereas 7/7 patients who remained PCR positive in this time period died (P = 0.006). In approximately 70% of evaluable patients, clinical relapse was preceded by recurrence of detectable MRD at time intervals of 3-18 months earlier and the recurrence of PCR positivity after a period of negativity was always followed by overt relapse. At relapse, the combined use of IgH and TCR delta probes reduced false negativity caused by clonal evolution to approximately 10%. This study shows that the evolution of PCR detectable MRD is an independent predictor of outcome.