This paper deals with the development of one of the basic elements of an expert computer system – the knowledge base – taking into consideration the particular applications. The situation is characterized by the fact that the knowledge of the given area is collected in the Guidelines for Management of Arterial Hypertension (2003), which summarizes the results of multiple and multinational studies on arterial hypertension. The formalization process of knowledge is time consuming and important in the creation of an expert system. At the first stage of formalization, all knowledge is gathered together from literature. Then decision trees are created. If the decision trees are correct, then they are transformed to the production rules. This paper describes an approach, which allows forming decision tree structure and production rules. The established knowledge base allows performing therapy selection for patients having arterial hypertension. Also the approach takes into consideration the situation, when multiple treatments should be selected.