The turbine blade in an atomic power plant may be fractured by fatigue, stress corrosion cracking and bad fitting. Especially, fatigue fracture is caused by low stress amplitude below the yielding stress. SEM fractography does not have striation, but AFM fractography does on the fatigue fractured surface of 12% Cr steel used for the turbine blade. Surface roughness $R_q$ measured by AFM is linearly related to the stress intensity factor range, ${Delta}K$, and is increased linearly according to the load range ${Delta}P$. Therefore, in this study, the loading condition applied to a turbine blade is predicted by the relation between the intersection of the ${Delta}K-R_q$ relation and load range ${Delta}P$.