1H-NMR (nuclear magnetic resonance) imaging is regularly proposed to non-invasively monitor cell therapy protocols. Prior to transplantation, cells must be loaded with an NMR contrast agent (CA). Most studies performed so far make use of superparamagnetic iron oxide particles (SPIOs), mainly for favorable detection sensitivity. However, in the case of labeled cell death, SPIO recapture by inflammatory cells might introduce severe bias. We investigated whether NMR signal changes induced by preloading with SPIOs or the low molecular weight gadolinium (Gd)-DTPA accurately monitored the outcome of transplanted cells in a murine model of acute immunologic rejection. CA-loaded human myoblasts were grafted in the tibialis anterior of C57BL/6 mice. NMR imaging was repeated regularly until 3 months post-transplantation. Label outcome was evaluated by the size of the labeled area and its relative contrast to surrounding tissue. In parallel, immunohistochemistry assessed the presence of human cells. Data analysis revealed that CA-induced signal changes did not strictly reflect the graft status. Gd-DTPA label disappeared rapidly yet with a 2-week delay compared with immunohistochemical evaluation. More problematically, SPIO label was still visible after 3 months, grossly overestimating cell survival (<1 week). SPIOs should be used with extreme caution to evaluate the presence of grafted cells in vivo and could hardly be recommended for the long-term monitoring of cell transplantation protocols.