Assisted driving is currently considered a key aspect for improving road safety, and automakers and OEMs are working to achieve higher levels of vehicle automation by introducing new technologies and Advanced Driver Assistance Systems (ADAS) in the circulating fleet. This trend requires test protocols for vehicle safety assessment to be frequently reviewed and updated, considering the latest advances in the state of the art regarding ADAS functions and systems. As of today, performance assessment programs (such as NCAP) mainly evaluate how an ADAS behaves in terms of crash avoidance in specific critical scenarios, which represent the most frequent crash constellations among real-world impacts. However, enhanced safety can be also obtained in case the impact is not avoided if a decrease in Injury Risk (IR) for the involved road users is achieved by ADAS intervention, compared to the case of no intervention.The purpose of this work is to propose an overall framework to draft or update test protocols for ADAS performance assessment based on real car-to-car impact observations, representative of impact scenarios in terms of both occurrence frequency and IR. First, the in-depth accident database IGLAD is analyzed to identify the most relevant car-to-car accident scenarios based on a relevance indicator, i.e., the risk level being the multiplication of the occurrence frequency and IR for a specific scenario. For each relevant scenario, a risk level-based strategy to identify one significant closing speed between vehicles for the tests is defined; the test collision speed for the two vehicles is determined analogously, and the risk level for each combination of speeds in a scenario represents the maximum achievable score by the ADAS if the collision is averted. Considering the well-established Euro NCAP framework as a relevant starting point for the definition of test protocols, two examples are highlighted regarding the proposal of a new test protocol and an update of an already existing one. Finally, a method is proposed for ADAS performance assessment if the impact is not avoided, scaling the maximum achievable score based on the IR reduction consequent to the ADAS intervention.