1. Glibenclamide, a blocker of ATP-sensitive potassium channels, has been shown to antagonize relaxin as a uterine relaxant in the rat in vivo but not in vitro. The aim, therefore, was to investigate whether the discrepancy between the two studies was a consequence of differences in (1) muscle layers, (2) hormonal conditions or (3) spasmogens utilized. Relaxin was compared with salbutamol and levcromakalim. 2. Relaxin was of similar potency as a uterine relaxant against oxytocin (0.2 mM)-induced spasm with tension measured in the circular or longitudinal muscle layers. Glibenclamide (10 microM) did not antagonize relaxin or salbutamol in these preparations but greatly antagonized levcromakalim (91-fold). Relaxin was a relaxant of tension activated by transmural electrical stimulation in uteri from rats that had been ovariectomized, although the maximal effect was only 30 +/- 15%, and in uteri from rats that had been treated with 17 beta-estradiol benzoate. Glibenclamide was not an antagonist of relaxin in the latter preparation but did antagonize levcromakalim (118-fold). Relaxin also inhibited spontaneous phasic tension development in uteri from ovariectomized rats but again was not antagonized by glibenclamide. 3. Because relaxin was not antagonized by glibenclamide under any of these various conditions, it would appear that the in vivo-in vitro discrepancy in the antagonism of relaxin by glibenclamide is not attributable to the effects of different muscle layers, hormonal conditions or spasmogens. It may be that the mechanism of action of relaxin or glibenclamide or both differs between in vivo and in vitro preparations.