Tofu production businesses are found in various regions, including Kolaka Regency. Tofu production businesses, which are generally on a household scale, usually depend heavily on achieving daily sales targets because tofu has a limited storage time. To avoid losses, tofu production must be carried out optimally. This research aims to optimize daily tofu production through implementing a Fuzzy inference system using the Mamdani method. This research depends on the availability of a dataset whose analysis process is experimental and numerical. The research results show that the fuzzy model can optimize daily tofu production from January to March. This claim is supported by the level of accuracy in the sufficient category, namely 77.7% when using RMSE, and the very good category when using MPE, namely 91.01%. Therefore, the fuzzy inference system model produced in this research can be used as an alternative for tofu industry businesses in determining daily tofu production.