Scatter of material properties is always an important factor in the assessment of integrity of structures of any kind. In nuclear power plants, fatigue is one of the major mechanisms which potentially leads to a failure of components. To evaluate scatter of fatigue strength in air and high-temperature water environment for austenitic stainless and ferritic steels (carbon steel and low-alloy steels), statistical analyses were carried out for data packages generated from fatigue test data obtained mainly in Japan. Best-fit equations for in-air data were derived using these data and they were used together with equations developed in Japan for environmental effect to obtain mean property of the materials. Distributions of the ratio of each test data to the mean property, in terms of the number of cycles to failure as well as strain amplitude, were statistically analyzed. In most cases, data scatter obeyed log-normal distribution quite well. Based on the regressions by log-normal distribution function, relations between design factors on them and failure probability were obtained for each material group and environment. It was found that the amounts of scatter of the in-water data were similar to that of the corresponding in-air data. Finally, design factors required to cope with material property variation are discussed.
