Weak infilled discontinuities commonly exist in rock masses, in which the infill degree of saturation largely influences the overalljoint shear behavior and ground stability. However, so far, research on the shear of infilled joints at unsaturated conditions is rare, especiallythose performed in numerical simulation. To the authors'knowledge, no attempts have been made to investigate the shear-induced variationsin unsaturated soil parameters, which are vital for proposing the infilled-joint constitutive models. For thefirst time, a series of constant watercontent direct shear tests on the unsaturated infilledjoint soil were conducted using the numerical software Fast Lagrangian Analysis ofContinua (FLAC)/Two-Phase Flow. Intrinsic soil-water retention and permeability models were updated in the FISH subroutine to considerporosity. Results highlight the disadvantage of the built-in models in FLAC. Initial infill saturation and other factors, including physical shearrate, joint roughness, infill thickness, and normal stress, all showed effects on the joint shear strength consistent with literature reports. Shear-induced variations in the mean values of Bishop effective stress and permeability of the infill layer were emphasized Abstract: Weak infilled discontinuities commonly exist in rock masses, where the infill degree of saturation largely influences the overall joint shear behaviour and ground stability. However so far, researches on the shear of infilled joints at unsaturated conditions are rare, especially those carried out in numerical simulation. Also to the authors' knowledge, no attempts have been made to investigate the shear-induced variations in unsaturated soil parameters that are vital for proposing the infilled-joint constitutive models. For the first time, a series of constant water content direct shear tests on the unsaturated infilled-joint soil were conducted using the numerical software FLAC/Two-Phase Flow. Intrinsic soil-water retention and permeability models were updated in FISH subroutine to consider porosity. Results highlight the disadvantage of the built-in models in FLAC. Initial infill saturation and other factors including physical shear rate, joint roughness, infill thickness and normal stress all showed effects on the joint shear strength consistent with literature reports. Shear-induced variations in the mean values of Bishop effective stress and permeability of the infill layer were emphasised.