D-optimal design and the desirability function were applied to optimize a self-microemulsifying drug delivery system (SMEDDS). The optimized key parameters were the following: 1) particle size of the dispersed emulsion, 2) solubility of the drug in the vehicle, and 3) the vehicle compatibility with the hard gelatin capsule. Three formulation variables, PEG200, a surfactant mixture, and an oil mixture, were included in the experimental design. The results of the mathematical analysis of the data demonstrated significant interactions among the formulation variables, and the desirability function was demonstrated to be a powerful tool to predict the optimal formulation for the explored system.