Treatment of pain with opioids is limited by their potential abuse liability. In an effort to develop analgesics without this side effect, a series of bivalent ligands containing a mu-opioid receptor agonist pharmacophore connected to a delta-opioid receptor antagonist pharmacophore through variable-length spacers (16-21 atoms) was synthesized. Members of this series [mu-opioid receptor (M)-delta-opioid receptor (D)-agonist (A)-antagonists (N): MDANs] are antinociceptive in the tail flick assay, but antinociceptive tolerance and physical dependence do not develop to ligands having spacers with 19-21 atoms. The current studies compared the rewarding properties of three bivalent ligands (MDAN-16, -19 and -21) and a mu-opioid receptor agonist (MA-19) to those of morphine in the conditioned place preference assay in mice after i.v. administration. Place preference developed to morphine and to MA-19, but not to the MDANs. The responses to MDAN-16 were highly variable, although place preference of borderline significance appeared to develop. Reinstatement was also evaluated after extinguishing morphine conditioned place preference; morphine and MA-19, but not the MDANs, reinstated morphine conditioned place preference. Taken together, these results suggest that the bivalents are less rewarding compared to morphine in opioid-naïve mice and do not induce reinstatement in previously morphine-preferring mice. The lack of a conditioned place preference response for MDAN-19 and -21, compared to the equivocal results with MDAN-16, suggests a minimum distance requirement between mu-opioid receptor and delta-opioid receptor recognition sites. This requirement may reflect the binding of MDAN-19 and -21 to mu-opioid receptor-delta-opioid receptor heterodimeric receptors that block reward but not antinociception.