The present paper is an attempt to optimize the condition of Siro spinning system for producing the elastic core-spun yarns by neural networks. Various factors such as feeding position of core yarn between two strands, distance between two strands, feeding angle of core yarn corresponding to strand axis, yarn twist level and tension ratio of core- elastic yarn affect significantly the physical and mechanical properties of core-spun yarn. Two cotton-polyester rovings (1.05 hank) and elastic yarn (40 dtex) were fed to drafting system and 20 Ne yarn was produced. To distribute the produced yarns in the appropriate classes, Kohonen net that is a competitive net was used. The learning rate and the final epoch were chosen on 0.2 and 105. First class is the class of the yarns that have the optimum properties.