This work concerns the probabilistic analysis of granular materials. More precisely, this work is devoted to the stochastic modeling of the granular geometric and constitutive microscale parameters associated with grain-pair interactions of an existing granular second gradient model [1]. Such an issue is addressed with a probabilistic methodology that relies on the Maximum Entropy (MaxEnt) principle from Information Theory [2]. After defining and improving the chosen granular second gradient continuum model, it is shown that the involved microscale parameters are statistically independent in micro-homogeneous and micro-isotropic materials. More precisely, the grain-pair distance between two consecutive grains is a uniformly distributed random variable and the specific microscale stiffness parameters are Gamma-distributed random variables. For illustration purposes, a micro-homogeneous, micro-isotropic, macro-homogeneous, and macro-isotropic 2D concrete plate subjected to an axial load is considered. A stochastic solver based on Monte Carlo (MC) numerical simulations is chosen whose convergence is studied. Finally, uncertainties propagation is discussed, and statistical fluctuations of the macro mechanical response are found to be nonnegligible. References [1] Placidi L., Berchiesi E., Misra A., et al., “Micromechanics-based elasto-plastic damage energy formulation for strain gradient solids with granular microstructure”, Continuum Mechanics and Thermodynamics, 33, 2213-2241, (2021). [2] Soize C., Uncertainty Quantification, Springer, Cham (2017).