Caffeine (10 mM)-induced relaxation of guinea-pig isolated trachealis was attenuated and converted to a small spasmogenic response on cooling to 22 degrees C. The relaxant response was restored on rewarming to 37 degrees C and was abolished by indomethacin (2.8 microM). Cooling to 22 degrees C in the presence of indomethacin revealed spasmogenic responses to caffeine which were abolished on rewarming to 37 degrees C. 2. Trachealis treated with indomethacin (2.8 microM) was repeatedly dosed with acetylcholine (ACh, 10 microM). Caffeine (1 or 10 mM), added as each ACh-induced spasm reached equilibrium, transiently augmented but then suppressed the spasm. On cooling from 37 degrees C to 12 degrees C, the increment in spasm evoked by caffeine increased relative to the spasm evoked by ACh. 3. Trachealis treated with indomethacin (2.8 microM) was repeatedly dosed with caffeine (10 mM). At 37 degrees C caffeine had little effect but it caused spasm when the tissue was cooled to 32 degrees C. Spasm amplitude increased as cooling progressed to 12 degrees C. Similar results were obtained with caffeine (1 mM). 4. At 37 degrees C, caffeine, enprofylline, 1,3,7,9-tetramethylxanthinium (TMX), theobromine, theophylline, xanthine and forskolin each caused concentration-dependent suppression of tracheal tone. Among the xanthine derivatives the rank order of potency was enprofylline greater than theophylline greater than caffeine greater than theobromine greater than xanthine greater than TMX. 5. In trachealis treated with indomethacin (2.8 microM) and maintained at 12 degrees C, the xanthines each caused concentration-dependent spasm. The rank order of potency was theobromine greater than or equal to theophylline greater than or equal to caffeine greater than or equal to enprofylline greater than xanthine greater than TMX. Forskolin was devoid of spasmogenic activity. 6. Trachealis treated with indomethacin (2.8 microM) and maintained at 12 degrees C, was repeatedly dosed with either caffeine (10 mM) or potassium chloride (KCl, 40 mM). Caffeine-induced spasm was attenuated in a Ca2+-free medium containing EGTA (2 mM), modestly at first but subsequently more profoundly. KCl did not evoke spasm at 12 degrees C but at 37 degrees C the KCl-induced spasm was virtually abolished at its first trail in the Ca2+-free, EGTA-containing medium. 7. It is concluded that caffeine, other alkylated xanthines and xanthine itself share a spasmogenic action in guinea-pig isolated trachealis which is best observed when the tissue is treated with indomethacin (2.8 microM) and maintained at 12 degrees C. The spasmogenic action represents the release of Ca2+ from intracellular sites of sequestration and may not depend on the intracellular accumulation of cyclic AMP. The rank order of spasmogenic potency of the xanthine derivatives differs markedly from their rank order of potency in suppressing the spontaneous tone of the trachealis observed at 370C. Since, at 12 degrees C, TMX is spasmogenic at concentrations identical to those causing relaxation at 37 degrees C, it is likely that TMX penetrates the cell. The relaxant effects of TMX do not, therefore, indicate that methylxanthine-induced relaxation is mediated by a receptor located on the external surface of the cell.