Safe-life methodologies such as the strain-life method require an accurate understanding of the notch root stress-strain evolution for accurate fatigue life estimation. Oversimplification of the local notch root stress-strain material behaviour can lead to inaccurate fatigue life estimation particularly for complex spectra containing a large number of cycles. In this paper, advanced nonlinear kinematic hardening models that simulate ratcheting behaviour are investigated in combination with large complex spectra. Preliminary results obtained from a crack growth program developed by the Defence Science and Technology Organisation (DSTO) called CGAP show that these advanced models provide only subtle improvements to fatigue life compared to that calculated using a less complex nonlinear kinematic hardening model. It is argued that the observed deficiencies in the advanced kinematic hardening models considered in this paper are the result of their inability to simulate mean stress relaxation.