An important feature of current expert system technology is the possibility of reasoning with real-time data generated outside of expert system itself. The purpose of this paper is to describe some issues of such reasoning. The description is supported by the experience gathered through the implementation of an expert system development tool, specifically designed for building real-time expert systems capable of reasoning with external data. Therefore the paper focuses on practical implementation algorithms and techniques, not merely on design . The issues covered include knowledge base organization, models of the reasoning process, reasoning speed, communication between the inference engine and the outside world in order to exchange real-time data, the independent operation of the inference engine and the outside processes, reasoning algorithms, timeliness of decisions made by the inference engine, and potential bottlenecks in the reasoning process.