Abstract: The most studies on ultrasonic nondestructive testing detection laid particular were qualitative, so the authors put forward a quantitative detection method of adhesive bond quality between metallic and nonmetallic components based on the neural networks.After considering the characteristic of the echo signals of ultrasonic detection to adhesive interface, the authors gained the characteristics that can reflect the adhesive bond quality by using the wavelet transformation.Then, the Elman neural network was used to process quantitative and classifiable recognization.The test results prove that the algorithm can recognize different echo signals quantitatively and effectively.