Protein purification has long been regarded as an art rather than a science. Trial and error is the widely used practice for a protein to be purified from the crude extract. In this study, we carried out a rational approach by analysis of the protein surface properties before setting out to do experiments. The target protein was recombinant human lactoferrin (rHLF), to be purified from milk of transgenic cow. We need to overcome two major problems. One is its possible co-precipitation with casein during initial separation; the other is the difficulty in separating its homologous counterpart, bovine lactoferrin (BLF). By calculation of the surface hydrophobicity, an initial separation step was decided by calcium precipitation, which removed casein but left rHLF in the supernatant. Then the average surface hydrophobicity and electric potential of rHLF were compared with those of BLF. There was a more significant difference in the electric potentials than the average surface hydrophobicity (ASH) between the homologous pairs. Therefore, the purification step should be favored by ion exchange chromatography (IEC) instead of hydrophobic interaction chromatography (HIC). Experiments were performed to verify the prediction. After removing casein, one step cationic ion exchange chromatography realized complete separation of rHLF from BLF with rHLF recovery up to 83%, while the resolution of HIC process was very limited due to the small difference in ASH between them. The laboratory process was then successfully scaled up to pilot-plant scale of 5001 milk of transgenic cow per batch. Average rHLF recovery of 79%and purity of 98.9% were attained for five batches. The purified rHLF displayed bioactivities as good as the natural human-resource lactoferrin standard. (C) 2015 Elsevier Ltd. All rights reserved.