The invention discloses a large-scale MIMO (Multiple Input Multiple Output) iterative detection method (MMSE-ANSE) with an adjustable convergence rate. The method comprises the following steps: preprocessing a channel matrix and a received signal vector to obtain a matched filter output vector and an MMSE (Minimum Mean Square Error) filter matrix; adjusting control parameters of an input convergence rate to ensure that an iterative matrix of a proposed iterative algorithm will be changed and is always related to lower triangular matrix elements of the MMSE filter matrix; setting an initial iteration solution as a zero vector, wherein the bigger the control parameters of the convergence rate are, the number of iterations is bigger, and the performance of an iterative detection algorithm is better; and the algorithm complexity of the detection algorithm is O (LM2). The large-scale MIMO iterative detection method disclosed by the invention is superior to a conventional Neumann series expansion-based detection method in the aspects of complexity and performance, especially under the condition that the communication environment is relatively poor.