To analyse the involvement of a possible numerical or qualitative stem cell defect in the development of sustained graft failure after autologous transplantation, we have determined the graft content of CD34+ nucleated cells, colony-forming cells and cobblestone area-forming cell subsets, as well as transplant ability to produce progenitors using the long-term culture colony-forming cell (LTC-CFC) assay. We evaluated material from the graft reference ampoules of 13 graft failure patients after bone marrow transplantation (BMT), four graft failure patients and four isolated thrombocytopenia patients after peripheral blood stem cell transplantation (PBSCT). We compared these data with those from six successfully engrafted BMT patients and 20 engrafted PBSCT patients respectively. In the BMT setting, the LTC-CFC 6-week assay represented a highly significant graft failure predictor. In the PBSCT setting, the total number of 2-week and 6-week LTC-CFCs transplanted per kg bodyweight (BW) showed the highest significant difference between the engrafted and the graft failure patients, as well as between the engrafted patients and the patients suffering from isolated thrombocytopenia after transplantation. These data show that the ability of a graft to generate progenitors in vitro rather than the number of primitive progenitors transplanted can have prognostic value for post-transplant haematological reconstitution.