The invention provides a mean square error based low-complexity MIMO interference channel receiver. According to the invention, the average value of each data flow MSE (Mean Square Error) is obtained in a statistics error range based on a channel statistics error model and considering that the BER performance mainly depends on the maximal data flow MSE at a high signal to noise ratio; and closed type sending and receiving vectors are subjected to irritation update by utilizing a Lagrange method and MMSE receiving and the parameter of a sending vector is determined through inner layer irritation optimization by taking the worst data flow MSE as the optimization object. Compared with an SOCP method, the invention has advantages of low calculation complexity and improvement of system BER performance.