The S-N curve regression parameters of structural steels were obtained by analyzing “Database on Fatigue Strength of Metallic Materials” published by the Society of Material Science, Japan. According to authors’ recent researches, it was found that S-N curves for structural steels can be estimated from the tensile strength of the respective steels. In such estimations, it was also found that the estimation accuracy could be improved by putting some classifications of the kinds of steels depending on the specimen fabrication process and fatigue testing conditions. Thus, the useful classification index (Ci) based on the S-N curve regression parameters and tensile strength σB is proposed to estimate the S-N curve for structural steels with the high accuracy. It was finally confirmed that Ci had the consistency with the “semi-logarithmic bilinear model”, and that the classification by using Ci has a sufficient validity on structural steels compiled in the above database. Another finding is that the specific number of stress cycles at the knee point of the S-N curve (Nw) has a clear dependence on both Ci and σB. That is, Nw tends to increases with decrease of Ci under the condition of same strength, while Nw tends to increases with increase of σB within the steel group giving the same level of Ci.
