Changes in cerebral blood flow (CBF) can be assessed directly with xenon clearance (XeC) or indirectly by measuring changes in middle cerebral artery blood velocity (Vmca) with transcranial Doppler (TCD). The aim of this study was to compare the changes in CBF and Vmca following caffeine ingestion. Nineteen patients (age 48-86, recovering from an acute stroke) and ten controls (age 52-85) were each studied twice. Bilateral measurements of CBF and Vmca were made before and after ingestion of 250 mg caffeine or matched placebo. The percentage change in CBF and Vmca after caffeine was calculated. Full results (CBF and Vmca) were obtained from 14 patients and 9 controls. There was no significant difference between patients and controls, so results were combined. Caffeine reduced CBF by 22% (95% confidence interval (CI) = 17% to 28%) and reduced Vmca by 13% (95% CI = 10% to 17%). The fall in Vmca was significantly less than that in CBF (p = 0.0016), showing that caffeine reduces mca diameter. Analysis based on Poiseuille flow in the arterioles suggests that caffeine reduced arteriole diameter by 5.9% (95% CI = 4.6% to 7.3%) and mca diameter by 4.3% (95% CI = 2.0% to 6.6%). TCD is being used as an alternative to XeC for assessing the effect of vasoconstrictors and vasodilators on CBF. This study has demonstrated that in mca diameter can be changed by the vasoactive agents, and that changes in Vmca do not necessarily reflect changes in CBF.