Electromagnetic source imaging techniques are usually limited by their low spatial resolution, even though these techniques have high temporal resolution. Our heuristic analysis shows that the spatial ambiguity of electromagnetic source localization arises from interference from other sources. In this paper, we suggest a new inverse solution based on the principle of spatial filtering to effectively suppress the interference from other sources, especially from the far sources. By means of this approach, functional MRI information can also be effectively integrated into the inverse solution to further improve spatial accuracy of source localization. Most importantly, the results of source localization by this approach are not significantly biased by incompatible fMRI information. Our simulations and experimental results using electroencephalography based on a realistic head model show that the Backus–Gilbert resolution spread function-constrained and functional MRI-guided spatial filtering suggested in this paper provide high spatial accuracy and resolution of source localization, even in the presence of multiple simultaneously active sources. Hum Brain Mapp, 2008. © 2007 Wiley-Liss, Inc.