With the advances in building automation systems (BASs), a large amount of building real-time operating data isbeing generated. This has lead building professionals to mine useful information from these data, and data mining algorithms have been proved to be powerful and effective tools. Supervisory control strategies, which have significant impacts on the operating efficiency, need to be actively assessed to keep building systems being operated properlyand efficiently. This paper hence proposes an automated recognition method to identify rule-based supervisory control strategies from operating BAS data. To accomplish this recognition target, a multiway decision tree was specifically designed, and it turned out to be suitable for this target. Finally, a case studyof operating the chiller plantis presented to show the effectiveness of this method and illustrate how to apply this method in practice.