We present a Monte Carlo analysis method for evaluating MRI-MEG/EEG co-registration techniques. The method estimates the error in co-registration as a function of position within the brain. Using this analysis technique, we demonstrate the limitations of conventional head-based fiducial point methods, and propose a new strategy utilising a dental bite-bar incorporating accurately machined fiducial markers. Results presented demonstrate the improved accuracy of MEG/EEG to MRI co-registration using the bite-bar.