Seizure is a phenomenon occurring in a large percentage of the population in every country of the world. It represents a distinctive indicator of an abnormality in the central nervous system. The main indicators in the EEG that predict a seizure are the presence of focal or multifocal receptive spikes or sharp waves (or both) and focal mono-rhythmic discharges. In this article we discuss the hardware and software that we developed for a system for automatic detection of these EEG signal characteristics at the beginning of an epileptic seizure. The patient is warned by the sounding of a special alarm. Such early detection helps to alert the patient to an impeding epileptic seizure and so allows him/her to take appropriate action to mitigate its effects.