MRI of superparamagnetic iron oxide (SPIO)‐labeled cells has become a valuable tool for studying the in vivo trafficking of transplanted cells. Cellular detection with MRI is generally considered to be orders of magnitude less sensitive than other techniques, such as positron emission tomography (PET), single photon emission‐computed tomography (SPECT), or optical fluorescence microscopy. However, an analytic description of the detection threshold for single SPIO‐labeled cells and the parameters that govern detection has not been adequately provided. In the present work, the detection threshold for single SPIO‐labeled cells and the effect of resolution and SNR were studied for a balanced steady‐state free precession (SSFP) sequence (3D‐FIESTA). Based on the results from both theoretical and experimental analyses, an expression that predicts the minimum detectable mass of SPIO (m_c) required to detect a single cell against a uniform signal background was derived: m_c = 5v/(K_fsl · SNR), where v is the voxel volume, SNR is the image signal‐to‐noise ratio, and K_fsl is an empirical constant measured to be 6.2 ± 0.5 × 10⁻⁵ μl/pgFe. Using this expression, it was shown that the sensitivity of MRI is not very different from that of PET, requiring femtomole quantities of SPIO iron for detection under typical micro‐imaging conditions (100 μm isotropic resolution, SNR = 60). The results of this work will aid in the design of cellular imaging experiments by defining the lower limit of SPIO labeling required for single cell detection at any given resolution and SNR. Magn Reson Med 53:312–320, 2005. © 2005 Wiley‐Liss, Inc.