The invention discloses an MIMO channel state information feedback method based on anti-fitting deep learning, and belongs to the field of communications. The method comprises the following steps that: firstly, an AOCN model is constructed, a channel matrix is divided into a real part and an imaginary part which are then input into an encoder of a user side, the encoder comprises a convolution layer and a full connection layer, data reaches a receiving end through a feedback link after being encoded, a decoder at the receiving end comprises an anti-fitting layer, a full connection layer, a RefineNet layer and a convolution layer, and finally a predicted channel matrix is output. After the AOCN model is constructed, offline training is performed on the model, model parameters are initialized firstly, the model is stored after error convergence, and finally, the trained and stored AOCN model is subjected to channel state information prediction online. According to the invention, the recovery precision of the information matrix can be further improved, the transmitting end of the system is ensured to obtain accurate channel state information, and the communication quality of the system is improved.