Water-alternating-gas (WAG) injection, both miscible and immiscible, is a widely used enhanced oil recovery method with over 80 field cases. Despite its prevalence, the numerical modeling of the physical processes involved remains poorly understood, and existing models often lack predictability. Part of the complexity stems from the component exchange between gas and oil and the hysteretic relative permeability effects. Thus, improving the reliability of numerical models requires the calibration of the equation of state (EOS) against phase behavior data from swelling/extraction and slim-tube tests, and the calibration of the three-phase relative permeability model against WAG coreflood experiments. This paper presents the results and interpretation of a complete set of two-phase and thee-phase displacement experiments on mixed-wet carbonate rocks. The three-phase WAG experiments were conducted on the same composite core at near-miscible reservoir condition; experiments differ in the injection order and length of their injection cycles. First, the two-phase water/oil and gas/oil displacement experiments and first cycles of WAG were used to estimate the two-phase relative permeabilities. Then, a history matching procedure over the full set of WAG cycles was carried out to tune the Larsen and Skauge WAG hysteresis model—namely the Land gas trapping parameter, the gas reduction exponent, the residual oil reduction factor and three-phase water relative permeability. The second part of this paper is dedicated to the value of information (VOI) analysis of the coreflood work program to assist the decision to proceed with a capital intensive WAG pilot at an offshore oilfield. Stochastic simulation of WAG injection using a fine scale sector model allowed to quantify the reduction in the range of uncertainty of key metrics—such as oil recovery, peak gas production and injectivity—linked with the additional SCAL information. The current study highlights the impact of the WAG injection sequence on the oil recovery and trapping mechanism. In addition, it is shown that the relative permeabilities and hysteresis model calibrated on one particular set of injection cycles fail to capture the WAG performance when the injection cycles are altered. Finally, the VOI methodology demonstrated the value enhancement from the coreflood work program.