Flotation kinetic models can be applied to describe the flotation process and to predict mineral recoveries. However, the size composition of the target minerals in the feed ore fluctuates considerably, resulting in insufficient accuracy with flotation kinetic models. There have been many studies that focus on the investigation of flotation kinetics with different particle sizes, while the optimization methods for flotation kinetic models based on particle size effects have not been reported. In this paper, flotation tests, optical microscope observations, and particle size analysis were used to identify the reasons for the decrease in accuracy of the flotation kinetic model due to changes in the composition of molybdenite particle size. Additionally, an optimization method for the flotation kinetic model was developed based on the particle size effect. The test results show that the accuracy of the flotation kinetic model for fixed particle size minerals is very high, but the predicted results for flotation recoveries of different particle size mineral mixtures have large deviations. The poor accuracy might be due to the autogenous carrier effect caused by the particle size composition fluctuating considerably. The optimization method for the flotation kinetic model is based on the particle size effect. The model can accurately describe the flotation process of molybdenite with different size compositions of molybdenite and predict the flotation recovery of molybdenite.