The increasing use of cultured human cells in clinical trials is highlighting the need for alternatives to media containing animal sera that are typically used to support these cultures. Perfused cultures of BM mononuclear cells (MNC) were used to evaluate animal sera alternatives with respect to the output of primitive, progenitor, and stromal cells. A serum level of 20% was optimal, and this could be provided by FBS alone or by a mixture of horse serum (HoS) and FBS, but not by HoS alone. Allogeneic human plasma (20%) supported half the level of cell, CFU-GM, and LTC-IC output as compared with animal sera-containing control. Significant donor-to-donor variability in human plasma was observed, but this was mitigated by pooling of plasma samples. Autologous and allogeneic human plasma performed equivalently. The use of autologous or allogeneic human serum was found to be equivalent to the use of human plasma, but all were inferior to animal sera. Animal sera supported typical stroma and cobblestone formation, whereas stroma in human serum cultures was less dense. Eight commercial serum-free media were tested and found to support MNC expansion to varying degrees, but none approached the performance of the animal serum-containing control, particularly with respect to stromal (i.e., CFU-F) support. In fact, when MNC were cultured in parallel with CD34-enriched cells, output (from MNC) was higher only in control medium, apparently because serum-free media reduced accessory cell effects. Because of these results, a new serum-free medium was developed for MNC cultures. This formulation outperformed all commercial serum-free media, resulting in cell and LTC-IC output equivalent to that of control. However, CFU-GM and CFU-F output were 66% and 9% of control, respectively. Precoating the culture surface with collagen increased CFU-F (and Thy-1+ cell) output to control levels, although CFU-GM output was still lower than control. The addition of either fibronectin or PDGF had no measurable effect, nor did the use of 5-100-fold greater concentrations of growth factor supplementation. The serum-free medium also increased CD41+ and CD61+ cell output to 150%-220% of control levels. The development of this new serum-free medium has potential for use in the perfused BM MNC culture systems currently in clinical trials to test the efficacy of expanded cells after cytoablative chemotherapy.