The operation of Enbridge’s pipeline system is continuously monitored by a leak detection system termed the Material Balance System (MBS). The MBS is not only a regulatory requirement; it helps minimize the financial and environmental damages that could result from an oil leak. A team of MBS experts provides 24/7 support to Enbridge’s control centre, which manages pipeline operations. When MBS predicts or detects any hydraulic abnormality in the pipeline, it alerts the operations’ staff. Though the operators do some basic analyses, such abnormal situations are usually referred to the support team. The idea of the intelligent alarm analyzer is to provide a tool to analyze and present the most probable cause of the alarm. The human knowledge of MBS alarm analysis is transferred into a soft knowledge base, which is then used in combination with the real-time data to analyze the alarm causes. Logic was developed to identify a group of alarms caused by a single reason, which was important from the performance point of view as a set of alarms can result from one event. In parallel, some basic rules were designed to implement the procedure for analyzing these alarms. A detailed study of some typical alarm scenarios was done to build an example knowledge base. The “event” triggers the alarm analyzer, which uses the real-time data and the knowledge base to go through an extensive series of logical evaluations and mathematical manipulations (called “tests”) to decide on the probable cause of the alarms. The next task was to combine the power of computational techniques and human understanding to implement possible scenarios. At this stage more complex and realistic rules have been developed to facilitate real-time alarm analyses. The paper will include an example scenario of how the tool could be applied to a production system. The alarm analysis system described in this paper is currently being tested for production use later this year. Once in production, additional tuning and refinement of the tool will be a continuing process.