A sensitive assay was developed for the detection of neuroblastoma cell contamination in CD34+ selected and unseparated peripheral blood stem cells (PBSC) used for autologous transplantation in stage 4 neuroblastoma patients. Specifically, we established a non-radioactive nested cDNA-PCR (nPCR) for detection of tyrosine hydroxylase (TH) gene expression combined with anti-disialoganglioside GD2 immunocytochemistry with the murine monoclonal antibody (MAb) 14G2a. Sensitivities of TH nPCR determined with a number of neuroblastoma cell lines and PBSCs correlated to cell line dependent basal TH gene expression levels and ranged from 1:10(4) to 1:10(6). The sensitivity obtained by immunocytochemistry was 1:10(5). We observed the highest PBSC contamination rate of 47% (18/38) among 38 PBSC specimens exclusively obtained from stage 4 neuroblastoma patients by using TH nPCR and GD2 immunocytochemistry in combination. Furthermore, a clinically applied purging method, CD34+ selection by immunoabsorption (CD34+ purity 42.4%), was used on 16 PBSCs. 10/16 (63%) preparations were contaminated prior to CD34+ selection and 56% (9/16) remained contaminated. A significant reduction of neuroblastoma cell contamination by CD34+ selection was not detectable, but the absolute amount of re-infused tumour cells was decreased due to 100-fold smaller cell counts of CD34+ selected grafts used for transplantation. 22 PBSC preparations were used for transplantation. A Kaplan-Meier analysis showed an event-free survival probability of 0.56 +/- 0.22 (n = 9) in the group with contaminated PBSCs versus 0.88 +/- 0.12 (n = 8) with no detectable neuroblastoma-cell contamination. Our data suggest that the combined use of TH nPCR and GD2 immunocytochemistry is optimal to detect contamination and monitor purging strategies.