Abstract: In order to improve the accuracy of cancer diagnosis,a model is proposed for cancer diagnosis,which based on the supervised Kohonen neural network.Adding an output layer on the basis of the Kohonen neural network,make the unsupervised neural network into supervised neural network-S_Kohonen which has the same network output number as the classification.The experimental results show that the supervised Kohonen neural network model is effective in clustering to the samples of the level of gene expression,so as to improve the accuracy and scientificity of cancer diagnosis,provide the certain reference value for the research on the treatment of cancer.