The RIS mechanism in austenitic and ferritic–martensitic (F–M) alloys is addressed by both atomic-scale modeling and experiment. Ab initio based modeling shows that the inverse Kirkendall mechanism for Cr in austenitic Ni-based systems is consistent with the observed Cr depletion, but that the result may be a balance of significant contributions from both vacancy and interstitial fluxes. In the F–M alloy system Kinetic Lattice Monte Carlo (KLMC) simulations of Cr segregation show that depending on the binding and diffusion energetics of Cr with vacancies, Cr diffusion by a vacancy mechanism can be either faster or slower than Fe. Experimental results show that RIS depends on the alloy composition and irradiation temperature, with grain boundary Cr enrichment occurring in the alloy with lower Cr (T91) and at lower temperature (400 °C), and depletion occurring in higher Cr alloys (HT9 and HCM12A) and at higher irradiation temperature (500 °C). Modeling results are entirely consistent with the expectations of an inverse Kirkendall mechanism determined by the relative rate of Cr to Fe diffusion.