A number of efforts have been made in the development of high-resolution EEG techniques, which attempt to map spatially distributed brain electrical activity with substantially improved spatial resolution. The objective of this study is to explore suitable spatial filters for inverse estimation of cortical potential imaging from the scalp electroencephalogram. The effects of incorporating statistically signal and noise information into inverse procedures were examined by computer simulations and experimental study. The parametric Weiner filter (PWF) with signal and noise covariance was applied to an inhomogeneous head model under various signal and noise conditions. The present simulation results suggest that, the PWF provides better cortical imaging results than the Tikhonov regularization under the condition of moderate and high correlation between signal and noise distributions. The proposed methods were applied to self-paced movement-related potentials (MRPs). The cortical potential maps estimated by means of PWF were well-localized in the premotor cortex, which is consistent with the hand motor representation.