Abstract: A probability-sorting based inspection method was proposed for chip type solder joints to improve the performance of automatic optical inspection on classification of solder joint, inspection accuracy and inspection speed. The algorithm of solder joint inspection was designed by analyzing the colorful images of solder joints, establishing model for each solder joint, defining inspection zone, extracting area of region, center of gravity and color gradient characters, and planning the inspection sequence. The experimental results illustrate that the proposed method could effectively detect and classify eight common types of chip solder joints with high inspection rate, which is useful for quality control in manufacturing process. And the inspection speed was obviously faster than other methods, which was lielplul to improve the efficiency of manufacturing process.