High-mobility wireless communication systems have attracted growing interests in recent years. For the deployment of these systems, one fundamental work is to build accurate and efficient channel models. In high-mobility scenarios, it has been shown that the standardized channel models, e.g., IMT-Advanced (IMT-A) multiple-input multiple-output (MIMO) channel model, provide noticeable longer stationary intervals than measured results and the wide-sense stationary (WSS) assumption may be violated. Thus, the non-stationarity should be introduced to the IMT-A MIMO channel model to mimic the channel characteristics more accurately without losing too much efficiency. In this paper, we analyze and compare the computational complexity of the original WSS and non-stationary IMT-A MIMO channel models. Both the number of real operations and simulation time are used as complexity metrics. Since introducing the non-stationarity to the IMT-A MIMO channel model causes extra computational complexity, some computation reduction methods are proposed to simplify the non-stationary IMT-A MIMO channel model while retaining an acceptable accuracy. Statistical properties including the temporal autocorrelation function, spatial cross-correlation function, and stationary interval are chosen as the accuracy metrics for verifications. It is shown that the tradeoff between the computational complexity and modeling accuracy can be achieved by using these proposed complexity reduction methods. P n Power of the n-th cluster. S Number of transmitter (Tx) antenna elements. s ASA Log-normal distributed random variable (RV) of angle spread of arrival (ASA). s ASD Log-normal distributed RV of angle spread of departure (ASD). s DS Log-normal distributed RV of delay spread (DS). s SF Log-normal distributed RV of shadow fading (SF). s K Log-normal distributed RV of Rician K-factor. T The number of time samples. U Number of receiver (Rx) antenna elements. v, θ v Speed and mobile direction of MS, respectively. v c , θ c Speed and mobile direction of mobile scatterer, respectively. φ n,m Angle of departure (AoD) related to the m-th (m = 1, • • • , M) ray within the n-th cluster. m,n Random initial phases related to the m-th ray within the n-th cluster. υ n,m Doppler frequency component related to the m-th ray within the n-th cluster. ϕ n,m Angle of arrival (AoA) related to the m-th ray within the n-th cluster. AMMAR GHAZAL received the B.Sc. degree in electronics and telecommunication engineer