The value of the tumour markers vanillylmandelic acid (VMA) and homovanillic acid (HVA) in urine (u) and serum (s), neurone-specific enolase (NSE), and lactate dehydrogenase (LDH) in the early prediction of relapse/progression in neuroblastoma is not known. We analysed the data of neuroblastoma patients who had successfully completed first-line treatment and had laboratory results available from their initial diagnosis and from relapse/progression (n=196). Patients' overall survival from relapse or progression was 21.5+/-4.2% (mean+/-standard deviation). At diagnosis, we found abnormal results in 75% for VMA and/or HVA (s), 92% for VMA and/or HVA (u), 90% for NSE, and 81% for LDH. We found a lower incidence of abnormal results at relapse or progression with 40% for VMA and/or HVA (s), 54% for HVA and/or VMA (u), 61% for NSE, and 48% for LDH. Sensitivity of all markers was higher for metastatic compared with local recurrence. NSE was the best, being able to detect 42% of the localised relapses, 77% of the combined local/metastatic relapses, and 69% of the metastatic recurrences. Relapse or progression in neuroblastoma cannot be detected reliably by monitoring tumour markers alone. Therefore, follow-up of neuroblastoma patients must include clinical assessment and imaging studies.