Background: Glioblastoma (GB) is the most malignant brain tumor in adults. It is characterized by angiogenesis and a high proliferative and invasive capacity. Standard therapy (surgery, radiotherapy and chemotherapy with temozolomide) is of limited efficacy. Innovative anticancer drugs targeting both tumor cells and angiogenesis are urgently required, together with effective systems for their delivery to the brain. We assessed the ability of human mesenchymal stromal cells (MSCs) to uptake the multikinase inhibitor, sorafenib (SFN), and to carry this drug to a brain tumor following intranasal administration. Method: MSCs were primed with SFN and drug content and release were quantified by analytical chemistry techniques. The ability of SFN-primed MSCs to inhibit the survival of the human U87MG GB cell line and endothelial cells was assessed in in vitro assays. These cells were then administered intranasally to nude mice bearing intracerebral U87MG xenografts. Their effect on tumor growth and angiogenesis was evaluated by magnetic resonance imaging and immunofluorescence analyses, and was compared with the intranasal administration of unprimed MSCs or SFN alone. Results: MSCs took up about 9 pg SFN per cell, with no effect on viability, and were able to release 60% of the primed drug. The cytostatic activity of the released SFN was entirely conserved, resulting in a significant inhibition of U87MG and endothelial cell survival in vitro. Two intranasal administrations of SFN-primed MSCs in U87MG-bearing mice resulted in lower levels of tumor angiogenesis than the injection of unprimed MSCs or SFN alone, but had no effect on tumor volume. We also observed an increase in the proportion of small intratumoral vessels in animals treated with unprimed MSCs; this effect being abolished if the MSCs were primed with SFN. Conclusion: We show the potential of MSCs to carry SFN to brain tumors following an intranasal administration. However, the therapeutic effect is modest probably due to the pro-tumorigenic properties of MSCs, which may limit the action of the released SFN. This calls into question the suitability of MSCs for use in GB therapy and renders it necessary to find methods guaranteeing the safety of this cellular vector after drug delivery.