Shunt insufficiency due to shunt occlusion or stenosis is frequent after TIPS (about 50% after one year). Controversially discussed is whether Doppler sonography is effective in detecting shunt stenosis or whether regular angiographies are required. The experience with a noninvasive method of surveillance primarily based on Doppler sonography is reported here. 58 patients (35 men, 23 women, mean age 55 years, range 33-82 years) were treated by TIPS because of complications of portal hypertension (43 x gastroesophageal bleeding, 14 x refractory ascites, 1 x venoocclusive disease). Liver cirrhosis (alcoholtoxic etiology in 63%) was present in 55 cases, according to Child-Pugh's classification 23 patients = A, 19 patients = B and 13 patients = C. Within a mean observation period of 14 months, Doppler sonography was performed in three months intervals, endoscopy in six months intervals and angiography only when shunt insufficiency was suspected by Doppler sonography and/or because of clinical events, e.g. recurrent bleeding. Immediately after TIPS, maximal flow velocity and flow volume in the portal vein increased by 116% and 115%, respectively. Three months later, a significant increase of portal vein diameter of about 15% was measured. Shunt flow was initially 2.700 ml/min (one week after TIPS) and decreased progressively by about 30% within the first nine months of follow-up. Correspondingly, angiographically proven shunt insufficiency was present in 22 patients (33 episodes). A total of twelve bleeding episodes recurred in seven patients (rebleeding rate of 16% after one year and 19% after two years). In the remaining 15 patients (68%; 21 episodes) shunt insufficiency could be corrected prior to complications because of detection by Doppler sonography (19x) and endoscopy (2x). Therefore, Doppler sonography is an effective diagnostic tool for the detection of shunt insufficiency and should be performed at three months intervals for at least 18 months. In this context it appears allowable to avoid routine angiographies.