In this paper, we propose a new robust content-based western music genre classification algorithm using multi-feature clustering (MFC) method combined with feature selection procedure. This paper focuses on the dependency problems of the classification result to different query patterns and query lengths which causes serious uncertainty of the system performance. In order to solve these problems, a new approach called MFC-SFSS based on k-means clustering is proposed. To verify the performance of the proposed method, several excerpts with variable duration were extracted from every other position in a same queried music file. Effectiveness of the system with MFC –SFSS and without MFC-SFSS is compared in terms of the classification results with k-NN decision rule. It is demonstrated that the use of MFC-SFSS significantly improves the system stability of musical genre classification with better accuracy.