During the operation of an industrial Fischer−Tropsch plant, fine catalyst particles are continuously generated in the slurry bed, which results in a series of problems, such as reduced flowability of liquid from the filter, reduction of unit processing capacity, and increased catalyst consumption. To solve these issues, a new classification method based on hydrocyclone was proposed to classify fine particles continuously. Cold model experiments using a mimic liquid−solid medium were conducted to verify the feasibility of this new method. The optimized structure and operating conditions were determined by studying the influences of structural parameters, operating conditions, and material physical properties systematically. Side-stream tests were then conducted to evaluate the actual classification performance. The results showed that the performance of the new classification method is primarily influenced by the medium temperature, i.e., the liquid viscosity. When slurry temperature is 200 °C, the coarse loss ratio can be reduced to less than 5% in a single-turn classification test and the fine removal ratio can be maintained at around 30%. With further increases in the temperature (i.e., at lower liquid viscosities), better classification performance can be expected. Finally, a prediction model was established, which can predict the whole classification performance of the new classification method. The model can be used to guide the process design and operation of this method.