A processing algorithm combined with wavelet analysis and information entropy theory is proposed to settle the problem of average and difference methods′ not being effective enough to evaluate the vibration events in phase-sensitive optical time-domain reflectometer (OTDR) vibration sensing system. Based on a well-developed discussion on the features of acquired signal and wavelet information entropy method, the effect the length of window and step on processing result is revealed. Meanwhile, an adoption of weighting algorithm is proposed for further optimization. Experiments prove that this method performs well in distinguishing noise from disturbance, and can recognize exterior interference in a rapid, accurate way as well as pinpoint the interference location, which can enhance the performance of phase-sensitive OTDR system.