In the 2009 version of the ASME BPV Code, a set of new design fatigue curves were proposed to cover the various steels of the code. These changes occurred in the wake of publications  showing that the mean air curve used to build the former ASME fatigue curve did not always represent accurately laboratory results.
The starting point for the methodology to build the design curve is the mean air curve obtained through laboratory testing: coefficients are then applied to the mean air curve in order to bridge the gap between experimental testing and reactor conditions.
These coefficients on the number of cycles and on the strain amplitude are equal to 12 and 2 respectively in the 2009 ASME BPV code, using the mean air curve proposal from NUREG/CR-6909 . Internationally, with the same mean air curve, other proposals have emerged and especially in France - where a consensus seems to be reached on the reduction of the coefficient on strain amplitude.
This paper provides statistical analyses of the experimental data obtained in France at high-cycle for austenitic stainless steels. It enables to bring arguments for the selection of a coefficient on strain amplitude in the French RCC-M code, where less scatter on the data is witnessed due to fewer material grades.