Current procedures for the genetic manipulation of hematopoietic stem cells are relatively inefficient due, in part, to a poor understanding of the conditions for ex vivo maintenance or expansion of stem cells. We report improvements in the retroviral transduction of human stem cells based on the SCID-repopulating cell (SRC) assay and analysis of Lin− CD34+CD38−cells as a surrogate measure of stem cell function. Based on our earlier study of the conditions required for ex vivo expansion of Lin−CD34+ CD38− cells and SRC, CD34+–enriched lineage–depleted umbilical cord blood cells were cultured for 2 to 6 days on fibronectin fragment in MGIN (MSCV-EGFP-Neo) retroviral supernatant (containing 1.5% fetal bovine serum) and IL-6, SCF, Flt-3 ligand, and G-CSF. Both CD34+CD38− cells (20.8%) and CFC (26.3%) were efficiently marked. When the bone marrow of engrafted NOD/SCID mice was examined, 75% (12/16) contained multilineage (myeloid and B lymphoid) EGFP+ human cells composing as much as 59% of the graft. Half of these mice received a limiting dose of SRC, suggesting that the marked cells were derived from a single transduced SRC. Surprisingly, these culture conditions produced a large expansion (166-fold) of cells with the CD34+CD38− phenotype (n = 20). However, there was no increase in SRC numbers, indicating dissociation between the CD34+CD38− phenotype and SRC function. The underlying mechanism involved apparent downregulation of CD38 expression within a population of cultured CD34+CD38+ cells that no longer contained any SRC function. These results suggest that the relationship between stem cell function and cell surface phenotype may not be reliable for cultured cells. (Blood. 2000;95:102-110)