The water transport mechanisms in organic matrices were reconsidered in two ways. Attention is focused on solubility rather than concentration in the analysis of equilibrium properties and a model in which polymer/water interactions govern, at least partially, diffusion is proposed. This model allows us to explain a certain number of experimental results, for instance: the fact that the number of sorbed water molecules is lower than the number of polar groups and that it increases pseudo parabolically with this latter; and the facts that diffusivity is globally a decreasing function of hydrophilicity; that its apparent activation energy is generally close to that for solubility, and that it tends to vary in the same sense with structure.