The multimodal transportation network includes a mix of inherently different modes. In addition to differences in price, range, and comfort of travel, these modes differ in mass and velocity, which correspond to different orders of magnitude in the kinetic energy carried. This discrepancy in kinetic energy affects both the level of protection of each mode, and the level of damage it can inflict on users of other modes. Unfortunately, accounting for both sides of a crash is often overlooked. While the quantities and variables of collected data continue to increase, the analyses conducted and the tools developed remain focused on the victims of crashes. The existing approach limits the ability to explore the underlying mechanism of traffic crashes since there are two sides to every crash. This manuscript proposes a framework for studying traffic safety that takes into account the interaction between all modes in a network. At the core of the framework is a square matrix, I. The rows and columns represent different modes such that element Iij is the number of injuries that were suffered by mode i, which were inflicted by mode j. The distinction between suffered and inflicted injuries is not related to the fault of the involved parties. The distinction lies in which of the two parties experienced the injury. For example, if two vehicles are involved in a crash that resulted in a single injury, the vehicle that experienced the injury is identified as the one that suffered the injury while the other vehicle is the one that inflicted the injury. If an injury is experienced in both vehicles then both vehicles suffered one injury and inflicted one injury. A relative vulnerability index can be calculated for specific mode-pairs, for individual modes, and for an entire geographical region. An empirical application using data from California reveals, amongst other things, that the relative vulnerability of pedestrian and bicyclist are orders of magnitude higher than motorized modes. Applying this methodology to different locations around the globe would provide insights the relative vulnerability of different modes under different mode-splits, different road designs, and different road user cultures.