HIV-1-derived lentivectors are promising for gene transfer into hematopoietic stem cells but require preclinical in vivo evaluation relevant to specific human diseases. Nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice accept human hematopoietic stem cell grafts, providing a unique opportunity for in vivo evaluation of therapies targeting human hematopoietic diseases. We demonstrate for the first time that hematopoietic stem cells from patients with X-linked chronic granulomatous disease (X-CGD) give rise to X-CGD-phenotype neutrophils in the NOD/SCID model that can be corrected using VSV-G-pseudotyped, 3rd-generation, self-inactivating (SIN) lentivector encoding gp91(phox). We transduced X-CGD patient-mobilized CD34(+) peripheral blood stem cells (CD34(+)PBSCs) with lentivector-gp91(phox) or amphotropic oncoretrovirus MFGS-gp91(phox) and evaluated correction ex vivo and in vivo in NOD/SCID mice. Only lentivector transduced CD34(+)PBSCs under ex vivo conditions nonpermissive for cell division, but both vectors performed best under conditions permissive for proliferation (multiple growth factors). Under the latter conditions, lentivector and MFGS achieved significant ex vivo correction of X-CGD CD34(+)PBSCs (18% and 54% of cells expressing gp91(phox), associated with 53% and 163% of normal superoxide production, respectively). However, lentivector, but not MFGS, achieved significant correction of human X-CGD neutrophils arising in vivo in NOD/SCID mice that underwent transplantation (20% and 2.4%, respectively). Thus, 3rd-generation SIN lentivector-gp91(phox) performs well as assessed in human X-CGD neutrophils differentiating in vivo, and our studies suggest that the NOD/SCID model is generally applicable for in vivo study of therapies evaluated in human blood cells expressing a specific disease phenotype.