Automatic interpretation of online measurements in clinical electrophysiological tests of cardiac function requires a modeling of electrical heart phenomena. The authors integrated a knowledge-based test support system into a real-time waveform recognition and measurement device. Knowledge is divided into three levels: object dictionary, interpretation rules, and textual explanations. Definitions are organized hierarchically as objects. The inferences about electrogenetic mechanisms of the arrhythmias and their pharmacological control are obtained in a quasi-real-time by means of heuristics that need only a couple of measurements, rather than complete statistics. Results and suggestions are explained interactively by texts and justified by electrophysiological models. A clinical example for atrial disrhythmias (atrial flutter) elucidates the knowledge-modeling approach.<<ETX>>