A controversy persists in autologous transplantation as to which source of progenitor cells, bone marrow (BM) or peripheral blood (PB), contains the lowest number of contaminating lymphoma cells, and how mobilization procedures affect these numbers. To accurately measure the number of non-Hodgkin's lymphoma (NHL) cells harboring the bcl-2/immunoglobulin H (IgH) rearrangement in progenitor cell grafts, we developed a nested quantitative competitive polymerase chain reaction assay (QC-PCR). DNA from lymph nodes of four patients with NHL were cloned into the pSK(+) vectors to generate four internal controls (ICs) (two with major breakpoint region [MBR] and two with minor cluster region [mcr] rearrangements). The kinetics of amplification of ICs paralleled those of bcl-2/IgH rearranged genomic DNA. When used in a QC-PCR assay, these ICs were accurate at a 0.2-log level and provided reproducible results, as shown by low intrarun and interrun variability. An excellent correlation between predicted and observed lymphoma cell content (r = .99) was observed over a range of at least 5 logs of rearranged cells. This approach was used to measure involvement by NHL cells at the time of progenitor cell harvest in 37 autologous transplant patients. The number of bcl-2/IgH rearranged cells in BM, PB, and mobilized PB (mPB) was found to vary from 1 to 1.1 x 10(5) per million cells. The number of lymphoma cells present in BM was significantly higher than in PB (P = .0001), with a median difference in lymphoma cell content between BM and PB of 0.48 log of cells (range, -0.7 to 5 logs). In contrast, we found no difference in the concentration of bcl-2/IgH rearranged cells present in BM versus PB progenitor cells mobilized with cyclophosphamide and granulocyte colony-stimulating factor (G-CSF) (mPB) (P = .57). In conclusion, the QC-PCR assay described in this study could measure accurately and reproducibly the number of bcl-2/IgH rearranged cells among normal cells. Differences in levels of contamination by lymphoma cells between BM and PB were of less than one log (10-fold), and no differences in lymphoma cell concentrations were observed between BM and mobilized PB. As more cells are usually infused with mPB than with BM grafts, mPB progenitor cell grafts may actually be associated with higher levels of contamination by lymphoma cells. Furthermore, this QC-PCR assay should provide an important tool to assess the prognostic impact of lymphoma cell burden both in progenitor cell grafts and in vivo.