Scalp surface Laplacian (SL) is widely used to enhance spatial resolution and sensitivity for electroencephalograph (EEG) recordings. In this paper, a radial-basis function (RBF) based surface Laplacian (RBFL) estimate is proposed for realistic head model, in which RBF is used as basis function to interpolate the surface of a head model and the surface potentials. The efficiency of RBFL was confirmed through a comparative study to the global realistic geometry spline Laplacian (GSL) with both simulation studies and human visual evoked potential. The simulations include both feasibility in a 3-concentric sphere head model and influence of head model noise and potential noise, effects of border source and sources separation distance in a realistic head model. All the comparisons show RBFL can provide comparatively better results than GSL and hence RBFL provide another efficient method for high-resolution EEG mapping.