A porous medium consisting of randomly packed spheres is generated and the fluid flow through such medium is solved numerically. The solution is used to generate both media consisting of mono- and poly-disperse spheres, and to predict and compare media properties: porosity, specific surface area and permeability. For poly-disperse media, five different distributions are used with spheres minimum to maximum sizes within an order in magnitude. In all cases, the number mean sphere diameter is kept constant. It is found for all three parameters that their values are not significantly affected with the sphere size distribution function and they remain close to the parameter values calculated for the mono-disperse medium. The porosity is close to 0.36 which is often reported porosity of the random pack of mono-disperse spheres, and the permeability of poly- to mono-disperse media increases up for one forth. This increase in the permeability goes on the expense of specific surface area as it is reduced for poly-disperse samples. If the number mean sphere diameter is used in Carman-Kozeny equation, the permeability of mono-disperse medium agrees very well with the numerical value, while the comparison worsens for poly-disperse packs. However, the comparison becomes much better if specific surface area is used in the permeability predictions. (C) 2019 Elsevier B.V. All rights reserved.