P T. Angsa Daya is a company engaged in the ceramic industry. The product studied in this study is SX type floor tiles with a size of 40x40cm2 which are produced on the production floor of Plant 3. The results of ceramic production at PT. Goose Daya is categorized into several types, namely ceramics of Quality 1 (KW1), KW2, KW3, KW4, and KW5. The determination of the category is carried out after the ceramics have been produced and inspected. Ceramics with the best quality will be categorized as KW1 ceramics, where these ceramics also have the highest selling price. PT. Angsa Daya has set a production target for KW1 ceramics which is 80% of the total production process. The research was conducted based on production data from June to August 2013, where the average production of SX KW1 type ceramics was 66.8%. The existence of a defect in the production results in a decrease in the production capacity of KW1 ceramics. The purpose of this study is to provide a proposal to improve the quality of SX type ceramics. The dominant type of defect that occurs is the Calibro defect, which is a type of defect caused by deviations from the ceramic size with the standard. The qualifying ceramic size deviation value for Quality KW1 is 0-1mm. From the Failure Mode and Effect Analysis (FMEA) table, it is found that the cause of the potential problem is the absence of a reference to the kiln machine setting. In this study, an experiment was conducted to obtain the optimal setting of the kiln machine using the Taguchi method. From the experimental results, it is found that the optimal Kiln engine setting is a temperature factor of 1100°C, the upper gas pressure is 55 Bar, the lower gas pressure is 55 Bar, and the wind pressure is 32 Bar. From the results of Analysis of Variance (ANOVA) the factors that have a significant influence are bottom gas pressure and wind pressure. After conducting a confirmation experiment, it was found that the average deviation of the ceramic size was 0.9mm. And by calculating the Taguchi Loss Function, the company's loss reduction before and after conducting the experiment was Rp. 304,667.67 with the assumption that the production cost was Rp. 1,000,000.00.