Optical time domain reflectometer (OTDR) is a very effective technique for distributed fiber sensing, in spite of its disadvantages including low spatial resolution, high cost and relatively noisy return. In order to improve the performance of OTDR, the concept of wavelet transform is introduced into OTDR in this paper. Wavelet transform is used to transform analysis from the time domain to the time- frequency domain, so not only the location of measurand can be detected more exactly, but also the character of frequency of the measurand can be analyzed. It has a wide use in distributed fiber sensors such as pressure, temperature, especially disturbance detection of one or more different physical measurand fiber sensor. In this paper, first the basic concept of OTDR is introduced and the fiber loss along the length of fiber is also analyzed. Second the principle of wavelet theory is described, it suggest that wavelet transform has a perfect character of localization and also a satisfying performance in the transient signal detection and signal decomposition from noisy background. At last results of analysis and experiment of wavelet transform in OTDR are presented by data and graph. The conclusion is that signals which is difficult to be analyzed and recognized in time or frequency domain can have a better resolution in time-frequency domain, it is important not only to backscattered OTDR but also other distributed fiber sensors.