In the present study, protein isoelectric point (pI), molecular weight (MW) and partitioning coefficient (K) were characterized for a mixture of proteins using a three-dimensional (3D) method which combined hydrophobic partitioning with two-dimensional (2D) electrophoresis. Protein pI and MW were obtained from 2D gels directly using the known pI gradients and the positions of MW standards. Protein partition coefficients were obtained by quantifying protein spot mass for the top and bottom phase gels so that mass ratios could be calculated for matched spots. The characterized three protein molecular properties were used to represent protein charge, size and surface hydrophobicity, which showed different degrees of influence on protein separation behaviors in ion-exchange chromatography (IEC).;Statistical models correlated the three characterized protein properties to retention times in cation-exchange chromatography (CEC) using partial least squares (PLS) regression. The resulting models fit well (R2=0.913 and 0.873 for SP and 15S, respectively) considering the limited property basis and the regression models were able to predict results for a small test set of proteins. The models showed that pI and MW correlated positively with CEC retention time, while the net influence of the partition coefficient depended on the base matrix type. This approach could be extended to host protein extracts to provide guidance for purification of recombinant proteins or choice of a suitable host for a particular recombinant protein.