The use of linguistic approximation enables knowledge to be represented in a more meaningful way and this is especially important in medical domain as it involves a lot of subjective decision making; Fuzzy Logic, introduced by Zadeh, has the ability to represent this imprecise expression. In this paper, an alternative approach of Fuzzy Logic in handling the approximate reasoning in expert systems will be described. The approach does not use the General Modus Ponen on the compositional rule of inference, but instead it uses a collection of rules to specify the properties of the inference (in making decision). An example of medical domain is described as its application.