In a previous paper (Watano et al., Powder Technol. 83, 55-60 (1995)), we developed a particle image probe and an image processing system to continuously monitor granule growth in fluidized bed granulation. This paper describes optimization of the image processing system in agitation fluidized bed granulation. First, the effects of operating parameters of the image processing system (e.g. the sensor location and the number of particles analyzed) on the results of granule measurements were investigated experimentally, and the optimum operating parameters were determined. Secondly, the performance and validity of the optimized system were investigated under various liquid how rates in actual agitation fluidized bed granulation. It was concluded that the optimized system could measure granule number median diameter and shape factor in fluidized bed granulation with high accuracy under various operating conditions.