Rock weathering has been investigated from atomic to global scales through the different but complementary approaches of mineralogy, petrography, geomorphology and geochemistry. The sequences of mineral reactions involved in the alteration process are now well known. They explain the global trend of weathering phenomena but do not account for the actual rock transformation dynamics. In particular, they ignore the intimate relation of the mineral reaction progress with the increase in connected porosity. At the hand specimen scale, heterogeneity is the rule: mineral reactions are controlled by local physicochemical conditions. Alteration processes depend largely on the rock microstructure properties. They proceed through nearly-closed, semi- and completely open microsystems which are interconnected by fractures or pores. Before being leached out by the solutions which flow in the large fractures (flux), the soluble elements migrate inside the connected porosity through chemical diffusion. The dissolution of the primary minerals is mediated through local gradients of chemical potential. With increasing alteration, the rock porosity increases, as does the length of the fluid passageways and their constrictivity and tortuosity. Consequently, the apparent diffusion coefficient for the most soluble elements decreases. The amplitude of the chemical potential gradients for the most soluble elements is reduced by the progressive coating of the reactive surfaces by clays and Fe oxyhydroxides. The residence time of these elements inside the weathered rock increases as alteration progresses; an effect enhanced by their temporary adsorption on the exchangeable sites of clays and Fe oxyhydroxides. Consequently, the weathering rate decreases with time. A possible new way to calculate weathering rates could be to measure the residence time of soluble elements inside the different microsystems during their migration towards the diluted solution which occurs in the large fractures.