The components of Dictamnus dasycarpus Turcz were tested for their vasorelaxing effect on the rat aorta, and fraxinellone and dictamine were shown to be effective vasorelaxants. In high K+ (60 mmol/l) medium, Ca2+ (0.03 to 3 mmol/l)-induced vasoconstriction was inhibited concentration-dependently by both agents. The IC50 for fraxinellone and dictamine were calculated to be about 25 mumol/l and 15 mumol/l (for Ca2+ concentration of 1 mmol/l), respectively. Cromakalim (0.2-10 mumol/l) relaxed aortic rings precontracted with 15 but not 60 mmol/l of K+. Fraxinellone and verapamil were more potent and effective in producing relaxation in 60 mmol/l than in 15 mmol/l K(+)-induced contraction. However, dictamine was more potent in producing relaxation in 15 mmol/l K(+)-induced contraction. Nifedipine (1 mumol/l), dictamine (100 mumol/l) and fraxinellone (100 mumol/l) relaxed the aortic contraction caused by KCl or Bay K 8644. The tonic contraction elicited by noradrenaline (NA, 3 mumol/l) was also relaxed by dictamine (500 mumol/l), but not by fraxinellone (500 mumol/l) in the nifedipine (1 mumol/l)-treated aorta. This relaxing effect of dictamine persisted in endothelium-denuded aorta. Glibenclamide (10 mumol/l) shifted the concentration-relaxation curve of cromakalim, but not that of dictamine, to the right in rat aortic rings precontracted with NA. Dictamine (500 mumol/l) did not affect tonic contraction of NA which are reduced by H-7 (1 mumol/l) in Ca(2+)-depleted medium. In conclusion, fraxinellone is a selective blocker of voltage-dependent Ca2+ channel, while dictamine relaxed the rat aorta by suppressing the Ca2+ influx through both voltage-dependent and receptor-operated Ca2+ channels.