Evaluating transceiver performance in the research and development stage is the basic function of over-the-air (OTA) testing. OTA testing is considered as the standard method for determining the radiated performance of wireless devices. The multiple-input-multiple-output (MIMO) technology used in wireless devices through the integration of multiple antennas is important for the current 4G long-term evolution (LTE) and the future fifth-generation (5G) wireless systems. OTA testing may be the optimal method for MIMO performance evaluation. During MIMO measurement, the transmitting power level of MIMO devices is set to 10 dB lower than the maximum transmitting power level. Because the transmitting power level can affect a device's physical temperature and, thus, proportionally increase the noise temperature and affect the throughput of the wireless equipment, the MIMO throughput of the device under test (DUT) determined with standard OTA measurement may not reflect real-world usage, and failure to consider the temperature effect of the DUT might contribute to measurement uncertainty. Here, to verify the temperature effect, we modified the radiated two-stage (RTS) procedure for MIMO systems to eliminate this uncertainty and quantify the temperature effect on DUT MIMO throughput. With this proposed method, the measured MIMO performance is a function of the temperature of the DUT, which is determined according to the transmitting power level of the DUT in its thermally stable condition. The proposed method further provides crucial information enabling engineers to optimize functional and EMC designs.