Microinjection of either baclofen (1.5 microgram) or morphine (2.5 microgram), in equimolar doses (14 mM), at sites located in the caudal periaqueductal gray (PAG) resulted in a delay in tail flick latency (analgesia). The relative analgesic potency of baclofen among caudal PAG sites, however, did not correlate with that of morphine. Application of either drug into the caudal aspect of the cerebral aqueduct also produced analgesia, but neither agent caused analgesia when applied at PAG sites rostral to the interaural line. Baclofen also produced analgesia when microinjected in the lower brain stem at sites lateral to the midline in or near the nucleus gigantocellularis, but did not produce analgesia when applied on the midline at sites located within or near the raphe magnus. Conversely, morphine produced analgesia when applied locally at midline sites but not at sites located lateral to the midline. These data suggest that the analgesia produced by systemic administration of baclofen and morphine involves activation of different neuronal substrates.