To determine the mechanism by which naltrexone (NTX) reduces oral ethanol-reinforced responding, opioid antagonists that show μ-, kappa-, or delta-selectivity were evaluated. Rhesus monkeys (n = 6) were given opportunities to respond and receive ethanol (1% or 2%) or water during daily 3-hr drinking sessions. Before some drinking sessions, the monkeys received intramuscular injections of saline or the following drugs: the μ-selective irreversible antagonist clocinnamox (CCAM), the kappa-selective long-lasting antagonist nor-binaltorphimine (nor-BNI), or the delta-selective antagonist naltrindole. Also, NTX was administered along with either CCAM or nor-BNI. When given alone, CCAM (0.1 mg/kg) had no effect on ethanol-reinforced responding. When NTX (0.32 mg/kg) was given with CCAM, responding maintained by ethanol was decreased. Nor-BNI (3 mg/kg) reduced ethanol-reinforced responding only on the day of injection. On subsequent days, when other studies report continued kappa-antagonism, responding maintained by ethanol returned to control levels. Also, NTX (0.32 mg/kg), administered in the presence of nor-BNI, was still able to reduce ethanol-reinforced responding. Naltrindole failed to alter responding maintained by ethanol. Because selective antagonism at the μ-, kappa-, or delta-receptor did not reduce ethanol-reinforced responding, NTX's ability to reduce ethanol consumption may not be mediated by these previously characterized opioid receptors. NTX may exert its effects through an uncharacterized opioid binding site or through a nonopioid mechanism.