P PT Argo Pantes, which operates in the textile industry, always strives to maintain and improve the quality of its products. Currently, the QC department at the spinning unit II of PT. ARGO PANTES is asking the production department to improve the quality of TC35-45 yarn (a blend of cotton (35%) and tetoron (65%)), based on one of its quality characteristics, namely nep count, because when compared to the production of spinning I and spinning III, the number of counted neps in the TC35-45 yarn of spinning II production is higher. Neps are a mass of tangled and very small fibers, where the fewer the number of counted neps, the better the quality of the cotton or yarn. Based on the cause and effect diagram, it is known that the process that most influences the characteristics of counted neps is the carding process, especially cotton carding. Measurements of the number of neps on two types of cotton carding machines were then carried out to determine the performance of the current quality characteristics of counted neps. The quality standards Kpl, Kp2 and S based on the number of neps in winding were converted into quality standards in carding, then a p control chart was created. From the p map, it can be seen that the proportion of quality Kp2 and S is still within the control limits. However, the company is not satisfied with the performance of the current calculated neps quality characteristics. Quality improvement is then carried out using the Taguchi method, one of the methods used in off-line quality control. Factors that influence the quality characteristics are sought using a cause and effect diagram. Due to limited facilities, for further research, only one type of machine is observed, namely the Meikin type. This one type is quite representative because based on the independent t test, it is known that the average proportion of Kp2 and S of both types of machines is not different. The characteristics that are the focus of the study are calculated neps which are divided into 3 categories, namely Kpl, Kp2 and S. The optimal setting sought is to minimize products that fall into the Kp2 and S categories. Based on the calculation results of the Taguchi experiment, the factors that significantly influence the average quality characteristics of calculated neps are the settings of the topflat gauge to the middle cylinder, topflat to the end cylinder, takerin to cylinder and dishplate to takerin. The best level for optimal setting is 0.004" for gauge doffer to cylinder setting, topflat to end cylinder 0.010", topflat to middle cylinder 0.007", takerin to cylinder 0.008", dishplate to takerin 0.011", topsheet 0.045" and doffer speed 16 rpm. Then a confirmation experiment was conducted. Loss function II calculated based on the results of the confirmation experiment, US$ 550.045, is smaller than loss function I calculated based on the initial setting of US$ 1,430.116. Savings occurred by 61.5385%. Improvements are also seen from the p control chart of the confirmation experiment compared to the initial p control chart, where p decreased from 0.4067 to 0.1667 and Cp increased from 0.5933 to 0.8333. From the results of data processing and analysis, it can be concluded that the optimal setting obtained using the Taguchi method is better than the initial production setting.