Résumé : High-dose therapy (HDT) and autologous transplantation prolongs remission duration and survival in multiple myeloma (MM), but relapse still occurs at a median of 2 years post-HDT. In order to investigate whether the number of residual tumour cells in the bone marrow (BM) after transplantation can predict the duration of response, a quantitative allele-specific oligonucleotide polymerase chain reaction (qASO-PCR) assay was used to measure tumour load in BM at 3-6 months post-HDT in 67 patients. The method of maximally selected log-rank statistics was used to test for the existence of a cut-off value in the BM tumour load data set. A cut-off value with respect to progression-free survival (PFS) was identified (P = 0.001). The estimated threshold for placing patients into a 'good' or 'bad' prognostic group was 0.015% (n = 22 and 38 respectively) with a median PFS of 64 months vs. 16. Multivariate analysis showed grouping by PCR result to be an independent prognostic factor for PFS (estimated hazard ratio after shrinkage, 3.91). This study identifies for the first time a threshold of the post-HDT tumour load with prognostic significance for PFS in MM. Quantitative molecular assessment thus may help to identify those patients who are in need of further treatment early after autologous transplantation.