For many years, spark-erosion processes have often been analyscd and controlled by real t,ime detection and evaluation of discharges in t h e gap. Mostly, normal sparks, short circuits, arcs and open circuits were distinguished; i.ies. particular arc type detections, etc. However, it turned o u t that some process phenomena could not fully be clarified and understood. The aim of this paper is to discuss a new developed "Pulse Discriminator (EDM-PD)" used for EDM process analysis and on-line control. The systems' principle i s based on continuous gap voltage tracking. Each appearing dischnrgc is analysed, and pulse characteristics are compared with preset values. It enables the discrimination of electric discharges in distinguished pulse typcs, all having a different impact on the process output, parameters as material removal, electrode wear, pollution of the gap, stability, etc. F u r t h ~r , the EDM-FD tracks pulse trains too. This contribut.ion is focusscd on the basic principles of the detect,ion device (EDM-PD). Approaches f o r metal removal and t.001 wear cvaluation on different. mat,eri-a1 combinat,ions are introduced. Finally, pulse train detection tochniqucs a'^e t.reskerl i~sing t.hs EDM-PD data. few systems had more advanced features 1 i k e discharge voltage noise de?t.ection faci li-