Molecular modeling of G-protein coupled receptors (GPCRs) remains a challenge due to the limited availability of structural information for the receptors. Molecular modeling approaches for melanocortin receptors (MCRs) fall into three categories: structurebased, ligand-based, and proteochemometric. Homology modeling combined with the information obtained from site-directed mutagenesis of receptors, recombined chimeric mutations of receptors and the structures of melanocortin type 4 receptor (MC4R) peptide ligands, has provided insights on detailed ligand-receptor interactions. Still, homology models based on the structures of bacteriorhodopsin (bR) or bovine rhodopsin as templates have not reached atomic level accuracy, making them unsuitable for rational drug design. On the other hand, availability of a large number of potent ligands of MCRs, especially those for the therapeutically important MC4R, has fueled ligand-based approaches, including automated pharmacophore query optimization and pharmacophorebased virtual screening. Proteochemometrics, a novel technology for the analysis of intermolecular interactions between ligand and receptor, has also shown great value in obtaining detailed informaton on molecular recognition and providing guidance to ligand design. In this review, the strengths and limitations of homology modeling, pharmacophore modeling and proteochemometrics modeling of MCRs are evaluated.