Approximately 0.3% of the worlds population is suffering from a focal epilepsy. In this paper the bioelectrical activity of the human brain in epilepsy will be analyzed with Volterra-Systems (VS) [1], whereas the kernels of different orders will be determined by using a RLS method of reduced computation complexity [2, 3]. The prediction gain for data segments obtained in presurgial evaluations will be given and criteria for a detection of distinct changes of the prediction coefficients will be proposed.