Brazil is the world's second largest producer of dicalcium phosphate, which is the main source of dietary phosphorus that complements the mineral salts contained in cattle, pig and poultry feed. The natural sources of dicalcium phosphate are not homogeneous and a final product classification process is required to ensure compliance with the physical specifications of the product shipped to clients. The objective of this work is to study a classification system for dicalcium phosphate using fluidization. A statistical analysis of the main experimental results using factorial planning and multiple regressions yielded parameters for the individual variables and the interactions among them. The model is validated by means of eight experiments performed in a pilot fluidized bed. The separation and recovery efficiencies obtained with the pilot system demonstrate the excellent technological application of the proposed dicalcium phosphate classification.