High-cycle thermal fatigue is a critical problem in nuclear power plants. To prevent crack initiation, Japan Society of Mechanical Engineers has issued a guideline for design, although growth analysis was not included. In this study, the feasibility of incorporating crack growth analysis into the design and integrity evaluation was investigated. Two characteristics of thermal fatigue loading were considered. The first was the effect of stress gradient in the depth direction. It was shown that the steep stress gradient near the surface significantly reduced the stress intensity factor (SIF) of deep cracks. Assuming that crack growth was arrested by small SIF values, it was judged possible to leave certain detected cracks unrepaired. Otherwise, the cracks should be removed regardless of their size. The other characteristic was the displacement controlled boundary condition. Through finite element analyses, it was revealed that the displacement controlled boundary condition reduced the SIF, and the magnitude of its reduction depended on the crack depth and boundary length. It was concluded that, under thermal fatigue loading, the cracks that were detected in the in-service inspection had already been arrested if they did not penetrate the wall thickness. It is effective to consider the crack arrest scenario for design and integrity assessment of cracked components under high-cycle thermal fatigue loading.