Abstract: Data mining has been effectively used in SMS classification which attracted the attention of researches in the last decades. The main objective of this study is to improve classification on an SMS junk using fuzzy-rough set approach with latent semantic analysis as feature selection which aims to reduce the error factor according to coverage factor reach to zero. The results are compared with the traditional fuzzy-rough set quick rules algorithm, decision tree and Naive Bays algorithms. The proposed method has satisfied better results for the classification rate according to the precision, F-measure, recall performance measures.