Some of the uncertainty regarding the use of cord blood (CB) in transplant settings includes the suspected relative rarity of hematopoietic stem cells (SC) in CB and the feasibility of incorporating a cell separation protocol to remove red blood cells, which may result in an unacceptable loss of SC. To address this uncertainty, we isolated a SC fraction by Percoll or Ficoll gradients from CB and peripheral blood (PB), which had been mobilized by chemotherapy. The frequencies of mononuclear cells (MNC), CD34+ cells, colony-forming units granulocyte-macrophage (CFU-GM), and the output of colony-forming cells (CFC) after five weeks in long-term culture (LTC) assay were then evaluated. The mean numbers of these cells per ml of CB sample before gradient separation were, respectively, 4.9 x 10(6), 13.8 x 10(4), 3.0 x 10(3) and 681 (n = 37). In the recovery phase of PB, these numbers were, respectively, 2.0 x 10(6), 14.9 x 10(4), 3.5 x 10(3) and 270 per ml of processed blood at apheresis (n = 35). After Percoll separation, the recovery rates of these cells were, respectively, 29%, 92%, 97% and 95% in CB, and 65%, 87%, 123% and 102% in PB. After Ficoll separation of CB, the rates were, respectively, 55%, 107%, 92% and 105%. These results suggest that CB contains an adequate number of more immature progenitors which can be retained after cell separation with Percoll or Ficoll, thereby making it feasible to incorporate a cell processing procedure into a CB transplant protocol. Percoll separation provided a greater enrichment of cells than Ficoll.