Spot welds are used extensively by automobile manufacturers as an efficient method for joining sheet steel. The quality of them can be tested non-destructively by using ultrasound. But when spot weld defect happens, the current ultrasonic inspection methods of weld quality are difficult to achieve an ideal result especially for the stick weld defect which is one of the most important types of spot weld defects in the automotive body. At first, this paper detailedly analyses the echo characteristics of ultrasonic curves which can reflect different spot welding joint defects. After echo characteristics qualitative analysis of different spot weld defects, a peak value marking algorithm is developed to identify the joint defects especially the stick weld defect rapidly and efficiently through selecting and confirming many optimal characteristic parameters. Finally, a lot of experiments are performed to verify the proposed methods. The results indicate that this fast identification method is credible and the identification rate can reach 95% in total test samples.