The timedomain analysis on waveform change of lightning electric field and the identification of discharge types are one of the most important works in the st udy of lightning detection system. The risetime, falltime, pulsewidth and a series of timedomain features of return strokes and narrow bipolar events we re given through analyzing the fast electric filed change data our system acquir ed. The risetime of the negative returnstroke fields had an average of 2.9μ s, its falltime and pulsewidth were 89μs, 15.4μs on average.The average risetime value of the narrow bipolar events was 1.7μs. The mean of the fallt ime value and pulsewidth value of narrow bipolar events were 2.1μs and 2.4μs , respectively. Some of the results were consistent with other researcher’s, bu t some differences existed which is caused by the different equipment used. Neve rtheless all the waveform features we got was still valuable for the identificat ion. Based on these characteristics, a set of software which could distinguish r eturn strokes and narrow bipolar events automatically from all of the lightning radiations events was designed. The classification results were tested by manual selection of data, and it proved a 90% correctness of identifying negative retu rn strokes and a desired accuracy of identifying positive return strokes and nar row bipolar events. As a preliminary research of the waveform classification of lightning discharge events, this method already has a practical value. It can be used to distinguish cloudtoground lightning, cloudtocloud lightning and narrow bipolar events in the new lightning detection system.