The overall goal of our research is an automatic, real-time and on-line monitoring system of artificial hearts. In this task, it is very important to automatically detect and classify abnormalities of the artificial heart control system and the recipient's circulatory system. The self-organizing map was applied to the pattern recognition of aortic pressure (AOP) which is considered to mostly represent the state of the circulatory system. The AOP signal data were fed to a Self-Organizing Map (SOM) beat by beat. During the unsupervised learning process the SOM units organize in such a way that similar AOP beat patterns were represented in particular areas of the SOM. The map location areas of the AOP signals in the different states of the circulatory system were also different. The results of visual examination revealed that the states of circulatory system were distinguished well by the map. It is expected that a map can be trained off-line with a large database and then used for on-line monitoring and analysis for artificial hearts.