Neural interfaces are aimed at creating an intimate contact with neural cells, either to stimulate the nerves or to record neural signals. This would allow the development of neurocontrolled artificial devices. The quality of these systems can be improved by increasing the selectivity of the neural interfaces used to stimulate or to record the activity of specific subsets of cells. Hence selectivity is of major importance for successful applications. The selectivity of these devices is a key issue and could be strongly influenced by the design of the microelectrode used. In this paper, a novel integrated FEM/Biophysical model based on anatomical and immunochemistry data providing guidelines for the design of more effective intraneural electrodes is used in studying the influence of active site shapes on the quality of stimulation and some preliminary results are also shown.