Use of umbilical cord blood (CB) for stem cell transplantation has a number of advantages, but a major disadvantage is the relatively low cell number available. Ex vivo cell expansion has been proposed to overcome this limitation, and this study therefore evaluated the use of perfusion culture systems for CB cell expansion. CB was cryopreserved using standard methods and the thawed unpurified cells were used to initiate small-scale cultures supplemented with PIXY321,flt-3 ligand, and erythropoietin in serum-containing medium. Twelve days of culture resulted in the optimal output from most CB samples. Frequent medium exchange led to significant increases in cell (93%), CFU-GM (82%) and LTC-IC (350%) output as compared with unfed cultures. As the inoculum density was increased from 7.5 x 10(4) per cm2 to 6.0 x 10(5) per cm2, the output of cells, CFU-GM, and LTC-IC increased. Cell and CFU-GM output reached a plateau at 6.0 x 10(5) per cm2, whereas LTC-IC output continued to increase up to 1.2 x 10(6) per cm2. Because the increase in culture output did not increase linearly with increasing inoculum density, expansion ratios were greatest at 1.5 x 10(5) per cm2 for cells (6.4-fold) and CFU-GM (192-fold). Despite the lack of adherent stroma, CB cultures expressed mRNA for many growth factors (G-CSF, GM-CSF, IL-1, IL-6, LIF, KL, FL, Tpo, TGF-beta, TNF-alpha, and MIP-1alpha) that were also found in bone marrow (BM) cultures, with the exception of IL-11 (found only in BM) and IL-3 (found in neither). Culture output was remarkably consistent from 10 CB samples (coefficient of variation 0.3) indicating that the procedure is robust and reproducible. Two commercial serum-free media were evaluated and found to support only approximately 25% of the culture output as compared with serum-containing medium. Implementation of optimal conditions in the clinical scale, automated cell production system (CPS) showed that the process scaled-up well, generating 1.7 x 10(7) CFU-GM (298-fold expansion) from 1.2 x 10(8) thawed viable nucleated CB cells (n = 3). The ability to generate >10(7) CFU-GM from <15 ml of CB within this closed, automated system without the need for extensive cell manipulations will enable clinical studies to test the safety and efficacy of expanded CB cells in the transplant setting.