Elucidation of the physiological role of the D3 receptor and its distribution in the brain using positron emission tomography (PET) is hampered by the lack of bioavailable subtype selective tracer ligands. To develop appropriate D3 radioligands, we designed an integrative procedure involving the elucidation of structural features determining D3 selectivity over both congeners D2 and D4 by comparative molecular analysis. Thus, we have successfully generated CoMFA and CoMSIA models based on the affinitiy differences of a series of 79 ligands representing a broad range of selectivities. These models yielded highly significant cross-validations (q2cv(D3/D2) = 0.86; q2cv(D3/D4) = 0.92) and excellent predictions of a 16-ligand test set (r2pred = 0.79-0.93). Exploiting this information, synthesis and receptor binding studies directed us to the fluorinated lead compounds 78 and 79, featuring subnanomolar D3 affinities and considerable selectivities over D2 and D4 and, subsequently, to the subtype selective PET tracers [18F]78 and [18F]79.