In the study of low cycle fatigue behavior, Manson-Coffin's linear relationships have been well recognized. However, some experimental results reveal a new evidence which is a divergence from Manson-Coffin's linear relationship between plastic strain range and number of cycles to failure under strain controlled cycling condition. This phenomena is closely related to the strength level above about 60 kg/mm2 and micro structure which has bainitic structure. For these steels linear relationship between total strain range and number of cycles to failure is applicable. The data in this investigation clarify that the method of predicting the low cycle fatigue strength proposed by Manson gives risky results. Therefore, it is desirable to predict the low cycle fatigue strength according to the following formula. For steels which satisfy Manson-Coffin's linear relationship : _??_ tr =0.715_??_ f 0.705· Nf - (0.558 nf +0.404) + (2.91σ u / E ) Nf - {0.056+1.29/ (σ u -26.3)} For steels which show a divergence from Manson-Coffin's linear relationship : in the low cycle range _??_ tr = (1.39×10-4σ u +0.715_??_ f 0.705) _??_ f - (1.08 nf +0.227) in the intermediate and high cycle range _??_ tr =1.39×10-4σ u · Nf - {0.056+1.29/ (σ u -26.3)}