Based on a request from the OECD Nuclear Energy Agency (NEA) Database Project FIRE in 2016, the recommendation was given by the PRISME 2 Program Review Group (PRG) that there should be a common Benchmark Exercise on a realistic cable fire scenario in an electri¬cal system being as far as reasonably practicable representative of a real cable fire event recorded in the FIRE Database, using information on electrical cable fires from the OECD/NEA PRISME (French acronym for “Fire Propagation in Elementary Multi-Room Scenarios”) Pro¬ject. The FIRE Database has clearly demonstrated the significance of fire events involving cables and shows that a majority of these events were either safety related or had the potential to impair nuclear safety. Both PRISME and PRISME 2 cable fire experiments have significantly increased the knowledge on cable fire behavior and investigated various types of cables im-plemented in nuclear power plants (NPPs) in member states. The major goal of this Bench-mark Exercise is to simulate a real cable fire scenario in order to assess and compare the capabilities of different types of fire simulation codes to model such a complex and realistic fire scenario. The strong interest of experts from regulators, technical safety organizations (TSOs) and licensees in predicting cable fires shows that such a Benchmark Exercise is a unique oppor-tunity for cross-cutting work between experts from the OECD/NEA FIRE and PRISME Pro-jects. Due to the high expert interest a decision was made to open the common PRISME 3 and FIRE Benchmark Exercise to other OECD/NEA CSNI (Committee on the Safety of Nuclear Installations) member countries. The real fire event selected for the Benchmark Exercise from the FIRE Database covering more than 500 fire events occurred in a heater bay of a NPP and involved two electrical cable trays loaded with PVC insulated cables. Since a numerical Benchmark Exercise on a real fire event is quite challenging, the following three-step methodology for conducting this Bench¬mark Exercise has been proposed: (1) a calibration phase, (2) a blind simulation of a PRISME cable fire experiment, and (3) the real fire event simulation. This methodology is based on the fact that a similar behavior is expected between the steps 2 and 3 making it possible to ex¬trapolate the error estimation.