The CoMFA methodology was applied to melatonin receptor ligands in order to establish quantitative structure-affinity relationships. One hundred thirty-three compounds were considered: they were either collected from literature or newly synthesized in order to gain information about the less explored positions. To this end, various melatonin derivatives were prepared and their affinity for quail optic tecta melatonin receptor was tested. Compounds were aligned on the putative active conformation of melatonin proposed by our previously reported pharmacophore search, and their relative affinities were calculated from the displacement of 2-[125I]-iodomelatonin on different tissues expressing aMT receptors. Compounds were grouped into three sets according to their topology. Subset A: melatonin-like compounds; subset B: N-acyl-2-amino-8-methoxytetralins and related compounds; subset C:N-acyl-phenylalkylamines and related compounds. CoMFA models were derived for each set, using the steric, electrostatic, and lipophilic fields as structural descriptors; the PLS analyses were characterized by good statistical parameters, taking into account the heterogeneity of the binding data, obtained with different experimental protocols. From the CoMFA model for the melatonin-like compounds, besides the well-known positive effect of 2-substitution, a low steric tolerance for substituents in 1, 6, and 7, and a negative effect of electron-rich 4-substituents were observed; the information provided by the newly synthesized compounds was essential for these results. Moreover, a comprehensive model for the 133 compounds, accounting for a common alignment and a common mode of interaction at the melatonin receptor, was derived (Q2 = 0.769, R2 = 0.905). This model validates our previously reported pharmacophore search and offers a clear depiction of the structure-affinity relationships for the melatonin receptor ligands.