To elucidate whether noninvasive arterial stimulation venous sampling (ASVS) is helpful for localizing insulinomas, calcium gluconate (0.02-0.025 mEq Ca2+/kg) was injected directly into the gastroduodenal, splenic and superior mesenteric arteries of six patients with episodic hypoglycemia. In all six patients, there was a greater than 26-fold increase in serum insulin levels in blood samples obtained from the hepatic vein after the infusion of calcium into the artery supplying the tumor. However, in four of the six patients, such an injection into an artery not supplying the tumor resulted in a greater than twofold increase in insulin concentration. Accurate localization of the insulinomas was verified at surgery in all patients. We believe that these false positive results were caused mainly by the influx of calcium via branches of intrapancreatic anastomoses. In order to minimize false positive results, we have recommended the following new criteria for ASVS: maximum insulin concentrations exceeding 150 microU/ml in blood samples obtained from the hepatic vein, and greater than twofold increases in insulin concentration. We concluded that noninvasive ASVS is helpful in determining the location of insulinomas.