Pithed rats were used to compare the abilities of vasopressin and NG-nitro-L-arginine methyl ester (L-NAME) to prevent the early (1 h after starting an endotoxin infusion) E. coli endotoxin-induced impairment of pressor responsiveness to noradrenaline, cirazoline, BHT 933 and to sympathetic stimulation (T8). L-NAME increased arterial blood pressure and augmented pressor responses to noradrenaline and to sympathetic nerve stimulation to a similar degree in control and endotoxin-treated rats. The response to the alpha 1-adrenoceptor agonist cirazoline was augmented by L-NAME in endotoxin-treated rats only, whereas the response to the alpha 2-adrenoceptor agonist BHT 933 was unaffected. Vasopressin (0.64 I.U. kg-1 h-1) prevented the hypotension that resulted from endotoxin administration and produced a similar increase in blood pressure to that produced by L-NAME. This dose of vasopressin also augmented pressor responses to noradrenaline and sympathetic nerve stimulation similarly in both control and endotoxin-treated rats. Sodium nitroprusside, in a dose that mimicked the degree of hypotension caused by endotoxin, also impaired pressor responsiveness to cirazoline; this impairment was prevented by co-infusion of vasopressin. Thus the effects of L-NAME in preventing the early phase of endotoxin-induced impairment of vascular responsiveness may be related to its hypertensive properties, due to inhibition of the constitutive form of nitric oxide synthase, rather than inhibition of endotoxin-induced nitric oxide synthase. These data suggest that early endotoxin-induced impairment of vascular reactivity probably involves factors other than nitric oxide. The well documented effect of endotoxin in inducing nitric oxide synthase probably explains the later, more sustained loss of vascular responsiveness.