In this paper, we described an approach to the automation of visual inspection of BGA solder joint defects of surface mounted components on printed circuit board by using neural network. Inherently, the BGA solder joints are located underneath its own package body, and this induces a difficulty of taking good image of the solder joints by using conventional imaging systems. To acquire the cross-sectional image of BGA sol-der joint, X-ray cross-sectional imaging method such as laminography and digital tomosynthesis has been cur-rently utilized. However, the cross-sectional image obtained by using laminography or DT methods, has inher-ent blurring effect and artifact. This problem has been a major obstacle to extract suitable features for classifi-cation. To solve this problem, a neural network based classification method is proposed int his paper. The per-formance of the proposed approach is tested on numerous samples of printed circuit boards and compared with that of human inspector. Experimental results reveal that the method provides satisfactory perform-ance and practical usefulness in BGA solder joint inspection.