OBJECTIVE--To report the first non-invasive assessment by transthoracic Doppler echocardiography of coronary blood flow in patients with aortic stenosis and of the effects of valve replacement. DESIGN--High frequency transthoracic Doppler echocardiography was used to examine resting phasic flow in the left anterior descending coronary artery before and after replacement of the aortic valve in awake, unsedated patients with pure aortic stenosis and normal coronary arteries. SETTING--A tertiary referral cardiothoracic centre. METHODS--Eleven patients with pure aortic stenosis and normal coronary arteries (six men, five women, mean (range) age 69 (50-82) years), were studied the day before and 1 week after replacement of the aortic valve. These patients were selected from a cohort of 15 due to ease of imaging of the left anterior descending coronary artery. Seven had a history of angina. Haemodynamics, peak transvalvar aortic gradient, left ventricular mass index, ventricular dimensions, and profiles of coronary flow velocity were measured. Profiles of coronary flow velocity were also measured in a control population of 10 normal subjects (five men, five women, mean (range) age 58 (34-66) years). RESULTS--The control population showed forward flow throughout systole, but reversed early systolic flow (mean velocity 20.6 (3.6) cm/s) was seen in six patients with aortic stenosis. Only three of these patients had a clinical history of angina. Peak and mean systolic and diastolic forward flow velocities were not significantly different in the control group and in patients with aortic stenosis. The time from the start of systole to the onset of forward systolic flow was significantly longer in patients with aortic stenosis than in the control population (185 (8.5) v 85 (10) ms, p < 0.01). The time from the onset of diastolic flow to peak diastolic velocity was also significantly longer in the aortic stenosis group (146 (16) v 74 (13) ms, p < 0.01). These abnormalities in profiles of coronary flow were reversed by replacement of the aortic valve. There was no correlation between changes in flow profiles in patients with aortic stenosis and preoperative clinical history, transvalvar gradient, left ventricular mass index, or ventricular dimensions. CONCLUSIONS--Coronary flow profiles in patients with aortic stenosis were characterised by reversed early systolic flow and delayed forward systolic flow and attainment of peak diastolic velocity. Reversal of these abnormalities by replacement of the aortic valve may reflect altered left ventricular and aortic haemodynamics and contribute to the relief of angina when left ventricular hypertrophy persists. Further studies may correlate abnormalities of coronary flow with preoperative clinical and haemodynamic state.