Transcranial Doppler-sonography (TCD) is a useful noninvasive technique for the measurement of intracranial arterial blood flow velocity. The present study evaluated the reproducibility of this method for repeated measurements of flow velocity in the anterior, middle and posterior cerebral, and basilar arteries. Thirty-six patients (23 males and 13 females) were studied. Three measurements were made by the same examiner on each patient with a time interval of 1 h between the first and the second examination, and about 24 h between the first and the third examination. The statistical evaluation for the reproducibility was performed by calculating the correlation coefficient (r). The results showed good reproducibility of TCD for all arteries both between the first and second and between the first and third measurements (r = 0.78-0.97; p less than 0.001), but when the middle cerebral artery mean velocity is low the posterior cerebral artery reproducibility is poorest (r = 0.42-0.54).