Water-alternating-gas (WAG) injection has demonstrated encouraging results for improving oil recovery. However, numerical simulation of three-phase flow and the associated hysteresis effects are not well-understood. Currently, modelling of hysteresis phenomena occurring in WAG cycles is based on updating two-phase relative permeabilities (krs) to account for three-phase characteristics, which is not a trivial task. In this work, a new assessment on the WAG-hysteresis model, which was originally developed for water-wet conditions, was carried out by automatic history-matching of two coreflood experiments in water-wet and mixed-wet conditions. Here in this work, it was attempted to estimate a set of more reliable two-phase krs within active saturation ranges of coreflood experiments performed under water-wet and mixed-wet conditions. Using the first cycle of the WAG experiments (first waterflood and the subsequent gas flood), the two-phase krs (oil/water and gas/oil) were estimated assuming minimal hysteresis occurs in the first cycle. For subsequent cycles, pertinent parameters of the WAG-hysteresis model are included in automatic history-matching process. The outcome of these historymatching exercises would indicate how the WAG-hysteresis model would perform under different wettability conditions using a more representative set of two-phase krs. Also, using these experiments, the ability of the WAG-hysteresis model in simulating the full cycles of WAG injections would be assessed. In other words, auto-history-matching of whole WAG coreflood experiments would enable integrated simulation of two and three-phase flow regions co-existing in cyclic WAG injection. The results indicate that history-matching the whole WAG experiment would lead to significantly improved simulation outcome. Values of the tuned Land trapping coefficient and the reduction factor for residual oil saturation were close to experimentally extracted information, which highlights the importance of two elements in evaluating WAG experiments; (i) including full-WAG experiments in history-matching and (ii) using a more representative set of two-phase krs. Therefore, employing our new methodology to estimate two-phase krs from first cycle of a WAG experiment could make the estimated unsteady-state krs more reliable for simulating hysteresis effects. Furthermore, the results show that the WAG-hysteresis model failed to reasonably predict the coreflood experiment performed at mixed-wet conditions. The failure can be attributed to two sources; 1) non-wet behaviour of water in mixed-wet systems, which should be reflected in three-phase water relative permeability curves and 2) over-simplifications assumptions embedded in the WAG-hysteresis model such as the linear Som reduction models. Hence, a series of modifications should be applied to change the formulations of the WAG-hysteresis model to develop a versatile model for mixed-wet conditions. (C) 2017 Elsevier Ltd. All rights reserved.