Naloxone-hydromorphone combinations were tested in participants trained to discriminate naloxone from placebo under a novel-response drug discrimination procedure while maintained on methadone. Naloxone alone produced dose-related increases in naloxone-appropriate responding, little or no "novel"-appropriate responding, and increases in opioid antagonist adjective ratings (n = 5). Hydromorphone alone produced dose-related increases in novel-appropriate responding, little or no naloxone-appropriate responding, and increases in opioid agonist adjective ratings (n = 6). When combined with naloxone, hydromorphone produced dose-related decreases in naloxone-appropriate responding and antagonist adjective ratings (n = 6). These findings are consistent with nonhuman data and suggest that this procedure may be useful as a human laboratory model of opioid withdrawal.