We studied spinal analgesic and antiallodynic effects of endomorphin-1 and endomorphin-2 administered i.t. in comparison with Tyr-D-Ala-Gly-MePhe-Gly-ol (DAMGO) or morphine, during acute, inflammatory and neuropathic pain in rats chronically implanted with intrathecal cannulas. Endomorphin-1 and endomorphin-2 (2.5, 5, 10 microg i.t.) increased the tail-flick latency and, to the lesser extent, the paw pressure latency. The range of potencies in both those models of acute pain was as follows: DAMGO > morphine = endomorphin-1 > endomorphin-2. In a model of inflammatory pain, the number of formalin-induced flinching episodes was decreased by endomorphin-1. The effect of endomorphin-2 was much less pronounced. Both DAMGO and morphine significantly inhibited the pain-related behavior evoked by formalin. In a neuropathic pain model (sciatic nerve crushing in rats), endomorphin-1 and -2 (5 microg i.t.) had a statistically significant effect on the tail-flick latency and on the cold-water tail flick latency. Morphine, 5 microg, was found to be ineffective. Endomorphin-1 and -2 (2.5 and 5 microg i.t.) dose-dependently antagonized allodynia. Those effects of endomorphins were antagonized in acute (30 microg), inflammatory (30 microg) and neuropathic pain models (60 microg) by cyprodime, a selective mu-opioid receptor antagonist. In conclusion, our results show a strong analgesic action of endomorphins at the spinal cord level. The most interesting finding is a strong, stronger than in the case of morphine, antiallodynic effect of endomorphins in rats subjected to sciatic nerve crushing, which suggests a possible use of these compounds in a very difficult therapy of neuropathic pain.