Combat fatigues are issued to military personnel with ready made clothes. Ready made combat fatigues should be fitted to various bodies of military personnel within given standard size. This paper develops a standard sizes selection method in order to increase the coverage rate and fitness for combat fatigues. The method utilizes a generalized learning vector quantization(GLVQ) algorithm that is one of cluster algorithm in neural networks techniques. The GLVQ moves the standard sizes from initial arbitrary sizes to next sizes in order to increase more coverage rate and fitness. Finally, when it cannot increase those, algorithm is terminated. The results of this method show more coverage rate and fitness than those of the other methods