An alarm system installed onboard a ship is often reported as a nuisance. It was even blamed as one of the causes of the accident. These facts draw the opposite function of the alarm system as the first layer of support to the operators during troubleshooting in the engine supervisory control. Additionally, the number of installed sensors on the machinery system increases with the automation’s implementation, likewise the number of alarms stored on the engine control consoles. Therefore, alarm management, a common practice in the onshore industry, becomes coherent also in maritime operations. It is beneficial for the operators who handle the alarms directly, and also for the onshore support staff, to grasp the onboard work situation. As the initial step of the alarm management, we conducted the alarm performance assessment with actual alarm data from the ocean-going vessel. The alarm data was retrieved in a data set containing the alarm name, alarm activation time, alarm deactivation time, and sensor reading value. This study developed several performance indexes based on modified methods ready on the literature: chattering index to categorize the nuisance alarms, similarity index between unique alarms, and similarity index between alarm floods. The actual data ship analysis shows the alarm performance assessment was able to discover several nuisance alarms and alarm floods. Thereafter, the practices in alarm management can be considered to minimize or eliminate these alarms, such as reconfiguring the alarm set-point, applying a delay time, and preparing the response strategy for alarm floods. Although alarm performance assessment only evaluated a small part of the human-machine interface, it provides added value in the age of digitalization and massive data communication. Alarm performance assessment can be a consideration for both onboard operators and shore management to maintain safe operations.