In batch crystallization processes, an operating temperature policy has a significant effect on the crystal size distribution (CSD) that determines final product quality. This work addresses the implementation of an on-line dynamic optimization integrated with a generic model control (GMC) strategy for improving the product quality of a seeded batch cooling crystallization. The on-line dynamic optimization strategy employed to determine the optimal operating temperature policy to minimize the nucleation of fine crystals is required to compensate errors regarding model uncertainties and unknown disturbances. Instead of assuming the perfect tracking of the optimal temperature profile, the GMC is applied to track the obtained optimal temperature policy. Simulation study on the batch cooling crystallization of potassium sulfate as a case study is presented. The results in cases of conventional linear cooling and optimal cooling with off-line and on-line computations are compared.