1. In the rabbit isolated aorta, atropine (3 x 10(-6) M-10(-4) M) inhibited contractile response to noradrenaline without affecting contraction to KCl. 2. In the presence of contraction to noradrenaline, atropine (3 x 10(-7) M-10(-4) M) caused concentration-dependent relaxation. Pretreatment with theophylline (10(-3) M) potentiated the relaxant action of atropine. Relaxation to atropine was not affected by the specific guanosine 3':5'-cyclic monophosphate phosphodiesterase inhibitor, M & B 22,948 (10(-4) M), tetraethylammonium (10 mM), indomethacin (10(-5) M), propranolol (10(-7) M), nifedipine (10(-6) M) or removal of the endothelium. 3. Relaxation to either atropine or prazosin was not affected by preincubation with prazosin and atropine, respectively. 4. In Ca(2+)-free medium containing EGTA and nifedipine, atropine (10(-7) M-10(-4) M) inhibited the residual noradrenaline response more than the subsequent Ca(2+)-induced contraction. Pretreatment with either theophylline (10(-3) M), forskolin (3 x 10(-7) M) or a low concentration of prazosin (3 x 10(-9) M) also inhibited the residual contraction to noradrenaline and Ca2+. The effect of combined treatment of atropine and any of these agents was much greater than with each individual agent. 5. Atropine (10(-6) M-10(-4) M) also inhibited increases in the level of inositol monophosphates (IP) in response to noradrenaline. Theophylline (10(-3) M) and a low concentration of prazosin (3 x 10(-9) M) also inhibited IP formation. Combined with atropine, the effect was much greater than with each of these agents individually. 6. Atropine did not affect adenosine 3':5'-cyclic monophosphate (cyclic AMP) levels in the aorta and also failed to displace specific [3H]-prazosin binding.7. These results suggest the possibility that smooth muscle relaxation to atropine may be due to the inhibition of phosphoinositide metabolism. The relaxation is not apparently due to an action of atropine on ax-adrenoceptors, or a change in the level of cyclic AMP.