Tests can reflect the students’ learning status. Because of the rising of cognitive psychology, so that researchers are beginning to pay attention to error responses in the course of test. At present, most discussions of problem-solving were focusing on the linear analysis of misconceptions, but very few on the similarity classify of the whole error responses. The traditional clustering method is difficult on finding the group type of misconceptions because of the characteristics of complex data in this research. There are 127 samples in this study which include 121 as general samples and 6 samples with learning disability that were identified by special educational identification group in school. Fuzzy C-means method was used to process the group clustering of misconceptions for obtaining the misconceptions group type and its educational significance.The result revealed that there were 3 clustering groups between all the samples’ misconceptions by using the clustering method FCM (Fuzzy C-meanings). The study also found that the error responses in the math test of most learning disabilities samples were clustered in the same clustering group. The result will provide teachers as a grouping reference in remedial and daily teaching.