[en] The paper points out how the accumulated knowledge of glass corrosion and the latest understanding of the glassy state need to be merged in order to improve the predictive power of existing corrosion models. In specific, two areas are identified which require substantial improvement. These are, firstly, an adequate description of the effects of glass composition, and, secondly, an explicit formulation of the pH dependence of the forward reaction rate. The former problem is solved by employing a constitutional approach to the Gibbs energy of the glass. The latter problem is tackled by taking into account the occupation of the glass surface by charged species (H⁺, OH^-, R⁺, etc.). (author)