Fishing is a very hazardous activity. Stability failures, caused many times by the crew lacking information related to the level of stability of the vessel, are among the most dangerous accidents affecting this fleet. A possible solution is offered by stability assessment systems that can help the skipper to identify potential risks and support his decision making process. The metacentric height is a key parameter for vessel stability and its real-time monitoring may be beneficial for alerting the crew about changes in stability. The paper proposes the design of a novel stability monitoring system that automatically detects changes in metacentric height based on estimates of the roll natural frequency solely using the measured roll angle. The core of the monitoring system is a combined estimation-detection system that exploits methods in advanced signal processing and statistical change detection to properly address issues of robustness. To analyze the monitoring performance, a nonlinear mathematical model of a stern trawler is used to generate roll motion time series in beam irregular waves of different peak period and significant wave height. Estimation and detection results obtained on the simulated data show a convincing ability of the monitoring system in discerning between safe and unsafe sailing conditions for most of the investigated cases.