Abstract: In order to establish a fatigue life prediction method for welded joints considering the influence of defects, the defect detection and quantitative characterization of 1Cr11Ni2W2MoV argon arc welded joints were conducted based on X-ray computed tomography (X-CT) technology, and a fatigue life prediction model for the joints was developed. Firstly, X-CT was used to detect microscopic defects in each micro-region of the weld and quantitatively characterize their morphology, size, and distribution pattern. The maximum characteristic size of welding defects was predicted through extreme value statistical analysis. The results show that the defect distribution in the heat-affected zone of the joint is relatively concentrated, and large-sized defects are mostly located near the surface of the specimen. The maximum value of the defect size φ_max is approximately 162 μm. On this basis, considering the variation of threshold values and closure parameters with crack size a during short crack propagation, a modified NASGRO model was established based on fatigue crack growth tests of the heat-affected zone in the welded joint. Combined with the defect characterization results, the fatigue life of the joint at different stress levels was calculated using the modified model. The comparison with the S-N curve results of fatigue tests shows that the model prediction results all fall within a ±2 scatter band, indicating high prediction accuracy. Furthermore, the modified model was used to expand the K-T diagram, and a defect-stress-life diagram of the welded joint was established, which can evaluate the fatigue life of the welded joint under different initial defect sizes.