The intradermal injection of adenosine produces a dose-dependent decrease in mechanical nociceptive threshold in the hindpaw of the rat that is not attenuated by elimination of indirect pathways for the production of hyperalgesia. Adenosine-induced hyperalgesia is mimicked by the A2-agonists, 5'-(N-ethyl)-carboxamido-adenosine and 2-phenylaminoadenosine but not by the A1-agonist, N6-cyclopentyladenosine and antagonized by the adenosine A2-receptor antagonist, PD 081360-0002 but not by the A1-antagonist, 1,3-dipropyl-8-(2-amino-4-chlorophenyl)xanthine. The latency to onset of adenosine and 2-phenylaminoadenosine hyperalgesia is similar to that produced by prostaglandin E2, a directly acting hyperalgesic agent but shorter than that produced by leukotriene B4, which acts indirectly. 2-Phenylaminoadenosine hyperalgesia is prolonged by rolipram, a phosphodiesterase inhibitor. Both 2-phenylaminoadenosine and prostaglandin E2 hyperalgesia are antagonized by the A1-agonist N6-cyclopentyladenosine and the mu-agonist, [D-Ala2, NMe-Phe4, Gly-ol]enkephalin. However, 1-acetyl-2-(8-chloro-10,11-dihydrodibenz[b,f]oxazepine-10-ca rbonyl) hydrazine, a prostaglandin-receptor antagonist, inhibits prostaglandin E2 (Taiwo and Levine, Brain Res. 458, 402-406, 1988) but not 2-phenylamino-adenosine hyperalgesia and PD 081360-0002, the adenosine receptor antagonist, inhibits 2-phenylamino-adenosine but not prostaglandin E2 hyperalgesia. These data suggest that adenosine is a directly acting agent that produces hyperalgesia by an action at the A2-receptor and that this hyperalgesia is mediated by the cAMP second messenger.