A methodology for extracting the domain knowledge from an existing three-tier web application and subsequent formulation of the platform independent model (PIM) is described. As it was devised during a reverse engineering process of an existing web application which needed to be reimplemented on a new platform using new technology, it focuses on the domain knowledge and business functions. It produces the business model and the hypertext model leaving the presentation model aside. The methodology is semi-automated — the generation of the activity diagrams and parts of the hypertext model must be in part performed by an analyst, preferably the one with some domain knowledge. As the paper is primarily aimed at practitioners, a case study illustrating the application of the presented method is included.