This study demonstrates a selectivity analysis using the GRID/CPCA strategy on four human cytochrome P450 2C homology models (CYP2C8, 2C9, 2C18, and 2C19). Although the four enzymes share more than 80% amino acid sequence identity, the substrate specificity differs. To investigate the selectivity of the enzymes and the amino acids that determine the specificity of each CYP2C enzyme, a selectivity analysis was made using GRID/CPCA. In the GRID calculations 10 probes were used covering hydrophobic, steric, and hydrogen bond acceptor and donor interactions. The selectivity analysis showed that the most important determinants of selectivity among the CYP2C models are the geometrical features of the active sites and the hydrophobic interactions. The selectivity analysis singled out CYP2C8 as the most different of the four CYP2C enzymes with amino acids with distinct properties in positions 114, 205, and 476 (Ser, Phe, and Ile, respectively) compared to the other enzymes. An inverse pharmacophore model for CYP2C9 was constructed from the selective regions, and the model agreed with the docking of diclofenac where the properties of the ligand overlapped with the pharmacophoric points in the model.