Critical infrastructures are important national assets for producing or distributing continuous flows of essential goods or services. When one critical infrastructure shuts down due to an external disruption, it can be expected that other critical infrastructures that need goods or services provided by the discontinued critical infrastructure will stop shortly, exacerbating the damage caused by the external disruption. Past research has categorized critical infrastructure interdependency (CII) into four relationship types and has proposed several methods to model CII and its effects. This paper presents a knowledge discovery process for CII that can be used to extract frequent patterns of critical infrastructure failure sequences triggered by external disruptions and/or CII. The knowledge discovery process, including integration of critical infrastructure failure records and transformation into the data format needed by the data mining algorithm, is described. The analysis results of real critical infrastructure failure records in certain areas of Taipei city are addressed, followed by discussion of a mitigation approach to stopping the possible future failure sequences of CII. Disaster mitigation officials can employ the proposed approach to explore CII and to design countermeasures when a disaster hits certain areas.