Abstract
This paper comes back on the accuracy of the surrogate-reaction method (SRM) historically used for neutron-induced average partial cross sections inference from measured surrogate-reaction probabilities. The SRM level of performance is examined in relation to a reasonably accurate reference calculation performed with the 𝒜𝒱𝒳𝒮ℱ-ℒ𝒩𝒢 code [1] through a challenging test case : the 240Pu* compound system. This paper argues on some ingredients of the reference calculation [2] and returns some hints about the failure now well-known of the neutron-induced γ average cross section inference. It shows also that in some special cases, the SRM can be poorly accurate also in terms of neutron-induced fission average cross section inference.