Image segmentation of fuzzy mid-wave infrared image is a challenging task according to the fuzziness of the image. Inspired by coordination network mechanism of biological immunity, a multilayer immune clustering neural network method is proposed, which based on imaging mechanism and clustering network statistical property of infrared image. First, by minimizing the between-class variance, fuzzy infrared image is segmented into three areas by preliminary immune clustering neural network, a bright area, a dark area and an unclear not dark area, then computes clustering network features of samples. Then, proceeds immune neural network clustering for samples in bright area and dark area based on clustering network features, thus each sample in unclear not dark area is divided into bright area or dark area. Experimental results show that the proposed algorithm can segment fuzzy infrared image efficiently. Compared with classical edge template and conventional region template methods, the multilayer immune clustering neural network method has better segmentation efficiency.