While recent reports have advocated the use of magnetic resonance imaging (MRI) to detect the effects of neuronal currents associated with human brain activity, only preliminary experimental data have been presented so far to demonstrate the feasibility of the method. Furthermore, it has not been adequately demonstrated that (1) MRI can separate neuronal current (NC) effects from other effects such as blood oxygen level-dependent (BOLD) contrast; (2) MRI has adequate sensitivity to detect NCs in vivo. In this work, we introduce a method that can separate slow (e.g., BOLD) processes from potential rapid (e.g., NC) processes and apply this method to investigate whether MRI allows detection of an NC response to a visual stimulus. MRI studies (n = 8) at 3.0 T using a sensitive multichannel detector showed insignificant effects related to NCs (averaged t < 0.05), in the presence of a highly significant BOLD signal (t = 6.15 +/- 0.90). In contrast, magnetoencephalography (MEG) experiments performed under similar conditions on the same subjects showed highly significant electrical activity (t = 7.90 +/- 2.28). It is concluded that, under the conditions used in this study, the sensitivity of MRI to detect evoked responses through NCs is at least an order of magnitude below that of BOLD-based functional MRI (fMRI) or MEG and too low to be practically useful.