Intraventricular administration of 5 or 20 micrograms of the cannabinoids WIN55,212-2 and CP-55,940 markedly reduced rat's responses to noxious thermal stimuli in the tail-flick test; no significant effect was found at 1 micrograms. The dose-response curves were steep and monotonic, the onset was rapid, and the effect lasted about an hour at the highest dose. In contrast to their antinociceptive actions, WIN55,212-2 and CP-55,940 failed to alter the latency of righting reflexes at the highest dose, suggesting that motor impairment did not cause the decreased responsiveness to the thermal stimulus. Finally, an assessment of the biodistribution of intraventricularly administered [3H]WIN55,212-2 in brain and spinal cord at the time of maximal antinociception revealed that the drug was confined to the brain. The levels of [3H]WIN55,212-2 found in S3-S4, the location of the spinal mechanisms for tail-flick, were below the limit of detectability. Together, these findings provide direct evidence that the antinociceptive effects of cannabinoids are mediated, at least in part, by their actions in the brain.