Morphine administered simultaneously to intracerebroventricular (i.c.v.) and intrathecal (i.t.) sites exhibits synergism, with the antinociceptive potency much greater than would be predicted from a simple addition of the potencies of the same dose administered to either site alone. This synergism was quantified in mice using both a fixed dose method, in which the morphine dose at one site was fixed while the AD50 (antinociceptive dose at 50% effectiveness) of morphine at the other site was determined; and a variable dose method, in which different doses of morphine were administered simultaneously to both sites at a fixed ratio, and the AD50 determined and compared to the AD50 at a single site alone. When animals were made tolerant to morphine by implantation of a 75-mg morphine pellet for 3 days, this synergism was eliminated, so that morphine administered simultaneously to i.c.v. and i.t. sites had an additive effect. However, administration of the peptide DynorphinA-(2-17) i.v. simultaneously to the test doses of morphine in morphine-tolerant animals resulted in a partial restoration of synergism. These results suggest that morphine-induced antinociception is highly dependent on an intact integrated central nervous system system and that the initial tolerance development is the result of a disruption of synergism between the central nervous system sites. Morphine tolerance results not from a reduced sensitivity to morphine at discrete central nervous system sites, but rather from a reduced synergistic interaction of morphine at spinal and supraspinal sites. In support of this conclusion, there was no tolerance observed in morphine-pelleted animals to morphine administered to i.c.v. or i.t. sites alone. DynorphinA-(2-17), a nonopioid peptide has previously been shown to enhance the antinociceptive potency of morphine in morphine-tolerant animals, appears to act by restoring this synergism.