In recent studies using a rat aortic balloon occlusion model, we have demonstrated that spinal grafting of rat or human neuronal precursors or human postmitotic hNT neurons leads to progressive amelioration of spasticity and rigidity and corresponding improvement in ambulatory function. In the present study, we characterized the optimal dosing regimen and safety profile of human spinal stem cells (HSSC) when grafted into the lumbar spinal cord segments of naive immunosuppressed minipigs. Gottingen-Minnesota minipigs (18-23 kg) were anesthetized with halothane, mounted into a spine-immobilization apparatus, and received five bilateral injections of HSSC delivered in 2, 4, 6, 8, or 10 μl of media targeted into L2-L5 central gray matter (lamina VII). The total number of delivered cells ranged between 2,500 and 100,000 per injection. Animals were immunosuppressed with Prograf® for the duration of study. After cell grafting, ambulatory function was monitored daily using a Tarlov's score. Sensory functions were assessed by mechanically evoked skin twitch test. Animals survived for 6-7 weeks. Three days before sacrifice animals received daily injections of bromodeoxyuridine (100 mg/kg; IV) and were then transcardially perfused with 4% paraformaldehyde. Th12-L6 spinal column was then dissected; the spinal cord was removed and scanned with MRI. Lumbar transverse spinal cord sections were then cut and stained with a combination of human-specific (hNUMA, hMOC, hNSE, hSYN) or nonspecific (DCX, MAP2, GABA, CHAT) antibodies. The total number of surviving cells was estimated using stereological quantification. During the first 12-24 h after cell grafting, a modest motor weakness was observed in three of eight animals but was no longer present at 4 days to 7 weeks. No sensory dysfunction was seen at any time point. Postmortem MRI scans revealed the presence of the individual grafts in the targeted spinal cord areas. Histological examination of spinal cord sections revealed the presence of hNUMA-immunoreactive grafted cells distributed between the base of the dorsal horn and the ventral horn. In all grafts intense hMOC, DCX, and hSYN immunoreactivity in grafted cells was seen. In addition, a rich axodendritic network of DCX-positive processes was identified extending 300-700 μm from the grafts. On average, 45% of hNUMA-positive neurons were GABA immunoreactive. Stereological analysis of hNUMA-positive cells showed an average of 2.5- to 3-fold increase in number of surviving cells compared with the number of injected cells. Analysis of spinal structural morphology showed that in animals injected with more than 50,000 cells/injection or volumes of injectate higher than 6 μl/injection there was tissue expansion and disruption of the local axodendritic network. Based on these data the safe total number of injected cells and volume of injectate were determined to be 30,000 cells delivered in ≤6 μl of media. These data demonstrate that highly reproducible delivery of a potential cell therapeutic candidate into spinal parenchyma can be achieved across a wide range of cell doses by direct intraspinal injections. The resulting grafts uniformly showed robust cell survival and progressive neuronal maturation.