Abstract: With growing wheat germplasm resources data, how to help breeding experts to obtain wheat germplasm efficiently and accurately has become an urgent issue．For this problem, a clustering algorithm-based wheat germplasm recommendation model was proposed．Wheat germplasm dataset was clustered using K-means to identify cluster centers of dataset．Cluster group was found to which breeding experts' germplasm data belonged, and then the nearest neighbor algorithm was used to derive wheat germplasm required by experts．Considering different contributions degree of wheat germplasm attribute features, gray weighted K-means clustering algorithm（GWK-means）was proposed．When calculating similarity of wheat germplasm by Euclidean distance, weight of wheat germplasm attributes were determined by grey correlation analysis, increasing distance between different clusters, improving accuracy and running speed of clustering algorithm, and providing a strong support for recommendation model．Experimental results on wheat germplasm dataset showed that average accuracy of the top 5 recommended wheat germplasm results and germplasm required by breeding experts reached more than 94%．