A two-stage design approach is introduced to determine the optimal feed rate, fed glucose concentration and fermentation time to maximize ethanol productivity by using Saccharomyces diastaticus LORRE 316. The first stage is applied to determine a kinetic model using batch time-series data. Two runs were carried out in the second stage to achieve the maximum ethanol production rate for the fed-batch fermentation process. The computational results from the kinetic model using batch and fed-batch time-series data are in better agreement with the fed-batch experiments than those from the kinetic model using batch time-series data. For cross-validation, an extra fed-batch experiment with constant feed rate was carried out for comparison with the optimal fed-batch result. According to the experimental results, this optimal approach was able to improve productivity by at least 12.56% compared with that using the constant feed rate.