During continuous operation of thermal systems, the performance of their components gradually deviates from the reference conditions due to degradation. As a result, the energy efficiency of these systems decreases substantially. In order to overcome this drawback, an effective predictive maintenance of the system can be implemented. However, this requires an appropriate quantitative assessment and identification of the impact of the malfunctioning components on the overall energy efficiency due to the complex interconnection among the components. Therefore, this work proposes a further development of the malfunction diagnosis of the components of thermal systems based on the exergy analysis. A case study involving a water-to-water heat pump operating at off-design operating mode was considered. The results obtained showed that the assessment of the conventional performance parameters and their deviations from the original values do not provide reliable information for the system performance evaluation. Moreover, this information was found to be misleading for the identification of the malfunctioning components as well as for the quantification of the faults. The application of the endogenous exergy destruction as a key parameter for the proposed malfunction diagnosis of a thermal system can successfully and effectively identify and quantify malfunctions in components, which affect both the remaining components and the overall system. The selected case study revealed that the proposed exergy-based malfunction diagnosis allows the investigated heat pump to save between 5 % and 8 % of electricity over the heating season at the studied conditions.