Atrial fibrillation (AF) has been reported as an independent risk factor of systemic thromboembolism. Almost half of the left atrial thrombi are located in the left atrial appendage (LAA). LAA function, reflected by LAA flow, thus has an influence on the potential of distal embolic complications. To identify factors other than atrial contraction that influence LAA flow during AF, transthoracic and transesophageal echocardiographic studies were performed on 130 patients. Seventy patients with nonrheumatic AF were divided into two groups with higher peak LAA outflow velocity (group 1) and lower peak LAA outflow velocity (group 2) at the ventricular systolic phase. Sixty patients with rheumatic AF were classified as group 3. Group 1 had a higher peak LAA outflow velocity than group 2 at both the ventricular systolic and diastolic phases. Group 2 had a higher peak LAA outflow at the ventricular diastolic phase than group 3 (18.9 +/- 8.0 vs. 11.8 +/- 7.5 cm/s, p < 0.001), whereas there was no significant difference in the peak LAA outflow at the ventricular systolic phase between the two groups (9.6 +/- 4.0 vs. 10.8 +/- 6.8 cm/s, p = NS). Group 3 was subdivided according to mitral valve area. Patients with severe mitral stenosis (mitral valve area < 1 cm2) had a significantly lower diastolic augmentation of LAA outflow velocity (difference of LAA outflow velocity between ventricle systole and diastole) than patients with mild to moderate stenosis (0.5 +/- 3.2 vs. 2.6 +/- 4.9 cm/s, p < 0.05). In conclusion, patients with rheumatic AF, especially those with severe mitral stenosis, have a lower diastolic augmentation of LAA outflow velocity. The lower diastolic augmentation of the LAA outflow velocity at the ventricular diastolic phase might result from interference with the suction effect of the left ventricular diastole by the stenotic mitral valve.