The paper investigates the use of probabilistic methodologies for assessing the performance of existing reinforced concrete (RC) buildings with limited ductility capacity retrofitted by means of buckling restrained braces (BRBs). In particular, the focus is related to the choice of adequate engineering demand parameters (EDPs) for these coupled systems that is one of the most important steps involved in the probabilistic framework. Probabilistic analyses require a high computational effort and hence, the chosen EDPs should be able to synthetically describe the structural behavior and, at the same time, to give an adequate description of the system demand. Global EDPs (i.e. the peak interstory drift) are usually adopted since they synthetically describe the response and since their use is more practicable during the analysis of complex models, however, previous studies highlighted that their use could leads to not conservative evaluations. On the other hand, the use of local component-specific EDPs, such as the strain demand at the most critical element sections although more cumbersome permits a more accurate evaluation of the system vulnerability. The system performance against earthquakes is commonly evaluated by using the seismic risk as main measure parameter and hence, it is used here to compare the use of global and component level EDPs.