The second part of this work is devoted to the study of holding time effects on the cyclic plastic behaviour of a martensitic steel tested at 823 K. Both relaxation and creep holding times of various durations were applied. The enhanced stress partitioning method presented in the first part [B. Fournier, M. Sauzay, C. Cae's, M. Mottot, Mater. Sci. Eng., submitted for publication] is used to evaluate the kinematic, isotropic and viscous parts of the cyclic stress. The bulk Young's modulus is found to vary significantly during cycling for creep-fatigue tests, which might be correlated to specific environmental interaction. The viscous stress measured at the end of the holding period tends to vanish as the holding time increases. The introduction of creep holding times enabled higher viscoplastic strains per cycle to be reached and allowed a larger range of strain rates to be studied. In all the cases tested, the observed softening effect is mainly due to the kinematic stress decrease. Nevertheless, even though the kinematic stress is always found to decrease with increasing accumulated viscoplastic strain, the initial magnitude of the creep-fatigue kinematic stress (measured at the end of the first holding period) can be either higher or lower than that of the corresponding pure-fatigue test. These effects of the holding period on the kinematic stress value can be related to the viscoplastic strain rate (and to the nature of the holding time: creep or relaxation). This dependency presents a maximum at intermediate strain rate, suggesting that two competing microstructural mechanisms control the magnitude of the kinematic stress. The enhanced stress partitioning method also enables the activation volume of both the creep and fatigue deformation mechanisms to be evaluated. The observed values are compared to those found in the literature