On the basis of a mu opioid receptor (MOR) homology model and the isosterism concept, three generations of 14-heteroaromatically substituted naltrexone derivatives were designed, synthesized, and evaluated as potential MOR-selective ligands. The first-generation ligands appeared to be MOR-selective, whereas the second and the third generation ones showed MOR/kappa opioid receptor (KOR) dual selectivity. Docking of ligands 2 (MOR selective) and 10 (MOR/KOR dual selective) to the three opioid receptor crystal structures revealed a nonconserved-residue-facilitated hydrogen-bonding network that could be responsible for their distinctive selectivity profiles. The MOR/KOR dual-selective ligand 10 showed no agonism and acted as a potent antagonist in the tail-flick assay. It also produced less severe opioid withdrawal symptoms than naloxone in morphine-dependent mice. In conclusion, ligand 10 may serve as a novel lead compound to develop MOR/KOR dual-selective ligands, which might possess unique therapeutic value for opioid addiction treatment.