In this paper, gap voltage and discharge current waveforms occurring during micro-EDM milling of micro-channels realized in a Si3N4-TiN workpiece are monitored and discriminated. The analysis is performed by implementing a design of experiment for the identification of the relationship existing among process parameters, machining performance and pulse type distribution. A pulse classification algorithm gathers gap voltage and current waveforms into four defined pulse types: short, arc, delayed and normal. The micro-channels are manufactured using an energy level corresponding to the finishing regime and varying pulse width, frequency and gap. The results show that material removal rate (MRR) benefits from the increase of normal and delayed pulses as expected. However, also arcs seems to increase MRR. Tool wear ratio (TWR) grows when normal pulses increases, whilst no particular influence is observed by delayed pulses. The peculiarity in TWR is found when arcs are considered: the values slightly decreases when arcs are more frequent, according to MRR behavior, but on the contrary with previous analysis done on different workpiece materials. This issue is currently under investigation.