In recent years there was a need for developing efficient NDE techniques for large area inspection. Conventional ultrasonic inspection techniques such as C-scan and B-scan are very time consuming because the transducer needs to be scanned over each point of structure under the test. Ultrasonically, Lamb wave is considered to be a candidate for large area inspection based on its capability of propagating long distances and its media-thickness dependent propagation properties. Unfortunately, Lamb wave inspection is complicated by the existence of at least two modes at any given frequency. The received signal generally contains more than one mode, and the proportions of the different modes present is modified by the mode conversion at defects. The modes are also highly dispersive in nature. As a result the peak detection and hence the flaw detection with these ultrasonic signals becomes highly complicated and the interpretation of the signals becomes very difficult and also leads to signal-to-noise problems. Consequently, a new kind of signal analysis is required to interpret the results of inspection and to determine if there is a defect or not. In order to extract the information on defects from the Lamb wave signal received by an ultrasonic transducer in such noisy environments, we use a method of analysis based on wavelet transforms. The detection of ultrasonic pulses using wavelet transforms is described and the robustness of this method is verified with testing on samples of adhesively bonded composite patch repair of cracked aluminum panels with simulated flaws.