Summary: The fuzzy logic inference method for decision making in expert systems of special class called situational is proposed. Decision inference in such systems is based on fuzzy classification of fuzzy input situation by comparison of its representation which describes the present (current) state of a controlled system. This situation is compared with the descriptions of standard fuzzy situations characterizing all possible states of a controlled system. Fuzzy situation inference of decision is enough simple for realization and in the same time has several advantages in comparison with the well known compositional inference rule. For fuzzy situation inference the architecture of coprocessor for IBM-compatible computers was developed which essentially uses natural parallelism of operations implemented over fuzzy sets. Due to this an acceptance of decisions making is possible in real time. Instrumental system for constructing expert systems on the base of fuzzy logic and fuzzy algorithms using fuzzy situational inferences is developed and a demonstration version of such expert system for IBM-compatible computers is created.