In order to improve the events analysis performance of optical time domain reflectometer (OTDR), locate the events accuracy and speedy, a new event analysis algorithm is particularly needed. Since OTDR signal contains nonstationary noise and the events can be seen as singularities, we presented an algorithm based on Empirical Mode Decomposition (EMD) and wavelet transform. Firstly using EMD to denoise the signal and then detecting the singularities by wavelet transform. This algorithm combines the advantage of EMD and wavelet transform. Experimental results show that this algorithm can detect and locate OTDR events effectively, and has great application value.