The invention discloses a semi-blind channel estimation method for an intelligent reflecting surface auxiliary communication system, which comprises the following steps that: firstly, a base station sends a corresponding intelligent reflecting surface reflection phase configuration signaling to an intelligent reflecting surface, wherein the intelligent reflecting surface is configured according to the signaling; a user sends a pilot symbol to the base station, and the base station performs initialization estimation on the cascade channel according to a received signal; then, the user sends a data symbol to the base station, and the data symbol is reflected to the base station through the intelligent reflecting surface; the base station firstly calculates a probability density function of a complete data set composed of a received signal and defect data, a log-likelihood function is solved for the probability density function, and by calculating cascade channel estimation when conditional expectation of the log-likelihood function is maximized, an accurate cascade channel estimation value is obtained through iteration. According to the invention, the cascade channel can be estimated by using fewer pilot signals, the pilot overhead of a large-scale intelligent reflector system is effectively reduced, and the communication rate of the system is improved.