In morphine-dependent mice, s.c. and i.t. administered naloxone produced withdrawal jumping (ED50 values were i.t. = s.c.) but i.c.v. administered naloxone failed to produce dose-dependent jumping. Peak times of jumping were earliest after i.t. administration of naloxone among the three administration routes. These results suggested that the spinal site was more sensitive to naloxone than the supraspinal site. Concomitant administration of naloxone i.c.v. and i.t. did not precipitate jumping. It was found that i.c.v. and i.t. injections of saline inhibited withdrawal jumping precipitated by s.c. administered naloxone and that the i.c.v. effect was more profound than the i.t. effect. I.c.v. injection of saline also delayed the peak time of withdrawal jumping precipitated by s.c. administered naloxone. These inhibitory effects of the injection procedures may explain the difficulty of i.c.v. administered naloxone and concomitant i.c.v. + i.t. administered naloxone to precipitate jumping, and may explain the difference in the ED50 values of naloxone and the time courses of jumping.