The objective of this study was to evaluate the effect of 2 independent formulation variables on the drug release from a novel doughnut-shaped minitablet (DSMT) in order to optimize formulations for intraocular drug delivery. Formulations were based on a 3² full-factorial design. The 2 independent variables were the concentration of Resomer (% wt/wt) and the type of Resomer grade (RG502, RG503, and RG504), respectively. The evaluated response was the drug release rate constant computed from a referenced marketed product and in vitro drug release data obtained at pH 7.4 in simulated vitreous humor. DSMT devices were prepared containing either of 2 model drugs, ganciclovir or foscarnet, using a Manesty F3 tableting press fitted with a novel central-rod, punch, and die setup. Dissolution data revealed biphasic drug release behavior with 55% to 60% drug released over 120 days. The inherent viscosity of the various Resomer grades and the concentration were significant to achieve optimum release rate constants. Using the resultant statistical relationships with the release rate constant as a response, the optimum formulation predicted for devices formulated with foscarnet was 70% wt/wt of Resomer RG504, while 92% wt/wt of Resomer RG503 was ideal for devices formulated with ganciclovir. The results of this study revealed that the full-factorial design was a suitable tool to predict an optimized formulation for prolonged intraocular drug delivery.