1. The bronchoconstriction of airway-perfused lungs and contraction of superfused tracheal spirals from guinea-pigs in response to adenosine were examined. 2. In lungs from untreated animals, adenosine had little effect unless the perfusion pressure was raised with carbachol (1.1 microM), when it caused a fall in perfusion pressure. However, if removed from guinea-pigs sensitized with ovalbumin (5 mg and 10 mg i.p. 14 and 12 days before use), adenosine was bronchoconstrictor, exerting bronchodilator effects only at high (1 mg) doses. The constrictor response to adenosine (300 micrograms) was significantly greater than that in lungs from untreated or sham-injected animals. 3. In superfused trachea from untreated animals, adenosine exerted only relaxant responses. In tissues from ovalbumin-sensitized guinea-pigs adenosine produced contractile responses, with relaxation appearing only at high (1 mg) doses. 4. Thus sensitization by antigen challenge revealed a bronchoconstrictor response of isolated airway preparations to adenosine. This is related to the clinical situation where only asthmatic subjects respond to adenosine by bronchoconstriction and suggests that the sensitization may destabilize inflammatory cells for mediator release by adenosine. 5. The response to a second exposure to adenosine was consistently reduced (lungs) or converted to a relaxation (trachea) indicating tachyphylaxis and consistent with a mediator release mechanism. 6. The P1-purinoceptor antagonist, 8-phenyltheophylline (3.9 microM), antagonized the relaxant responses to higher doses of adenosine. However, it did not affect the contractile responses to lower doses of adenosine. Whether this is due to P,-purinoceptors not being involved in the contractile response, or whether preferential blockade of the relaxant response leaves the contraction unopposed and apparently unblocked, remains to be established.