The ultimate therapy for ischemic stroke is restoration of blood supply in the ischemic region and regeneration of lost neural cells. This might be achieved by transplanting cells that differentiate into vascular or neuronal cell types, or secrete trophic factors that enhance self-renewal, recruitment, long-term survival and functional integration of endogenous stem/progenitor cells. Experimental stroke models have been developed to determine potential beneficial effect of stem/progenitor cell based therapies. To follow the fate of grafted cells in vivo, a number of non-invasive imaging approaches have been developed. Magnetic Resonance Imaging (MRI) is a high resolution, clinically relevant method allowing in vivo monitoring of cells labeled with contrast agents. In this study, labeling efficiency of 3 different stem cell populations (mouse Embryonic Stem Cells, rat Multipotent Adult Progenitor Cells and mouse Mesenchymal Stem Cells) with three different (ultra) small superparamagnetic iron oxide (U)SPIOs particles (Resovist(R), Endore(R), Sinerem(R)) was compared. Labeling efficiency with Resovist(R) and Endorem(R) differed significantly between the different stem cells. Labeling with (U)SPIOs in the range that allows detection of cells by in vivo MRI, did not affect differentiation of stem cells when labeled with concentrations of particles needed for MRI-based visualization. Finally, we demonstrated that labeled rMAPC could be detected in vivo and that labeling did not interfere with their migration. We conclude that successful use of (U)SPIOs for MRI based visualization will require assessment of the optimal (U)SPIO for each individual (stem) cell population to ensure the most sensitive detection without associated toxicity.