Abstract: Aiming at the problem of near surface dead zones and defects automatic identification in ultrasonic TOFD technique, a automatic identification technology of near surface defects is proposed based on through wave of ultrasonic TOFD and neural network. Several key points in the part of through wave of testing signal are extracted and relationship between the amplitude distribution of key points and depth of near-surface defect is analyzed. The characteristic numbers of amplitude distribution which can be used to test near-surface defect are obtained. Moreover, the characteristic numbers can be used to defects recognition and classification in BP neural network. The experimental results showed that this technique can be used for accurate and effective classification and automatic identification.