Alternatives to vapour compression refrigeration cycles are needed to eliminate the use of pollution intensive refrigerants. Magnetic refrigeration (MR) exploits the magnetocaloric effect to create a refrigeration process using environmentally benign fluids and materials. Current research efforts in the field of MR focus on optimizing the active magnetic regenerator (AMR), a unique component of a magnetic refrigerator which houses the magnetocaloric material (MCM). For MR to realize its full potential as a vapour compression cycle alternative the AMR must be optimized to lower cost and improve performance. An issue in the research field of MR is the difficulty of comparing the performance of different devices in various research facilities around the globe. The standard performance metric in the field of refrigeration is the COP, which is measured on vapour compression cycles using a standard measurement procedure which is not compatible with MR devices. Additionally, the COP captures overall device performance – MR research centers around the optimization of the AMR, so a performance metric which isolates this key component and neglects inefficiencies from motors and frictional losses is more useful. Thus, the objective of this work is to define a performance metric which captures the key research targets of MR and can be compared across devices with alternate orientations, cooling capacities and AMR compositions. The COP_AMR performance metric has been shown to isolate the AMR in previous work, although there are challenges measuring this value on devices with a Halbach array magnetization method. This work investigates calculating the COPAMR by measuring the heat rejection from the hot side heat exchanger (HEX) using a heat flux sensor (HFS). It is shown that the heat rejection from the hot HEX of a MR can be characterized as a function of device temperatures and measured heat flux from the sensor, despite the transient (oscillating) flow occurring in the HEX. By calibrating the HFS/HEX system with a resistive heating pad outputting a known heat rejection, it was possible to correlate the sensed heat to the overall heat rejection. Experiments performed on a testbench HEX device yielded relationships which can be used to calculate the total heat rejection given the sensed heat, cooling fluid temperature, and HFS temperature. Finally, the various errors of the procedures are examined, and the proposed method to retrofit MR devices with COP_AMR measurement capabilities is stated.