Pawlak’s Rough Sets Theory is one of many mathematical approaches to handle imprecision and uncertainty. The main advantage of the theory over other techniques is that it does not need any preliminary or additional information about analyzed data. This feature of rough set theory favors its usage in decision systems where new relations among data must be uncovered. In this paper we use data from a medical data set containing information about heart diseases and applied drugs to construct a decision system, test its classification accuracy and propose a methodology to improve an accurateness and a testability of generated “if-then” decision rules.