In vitro studies have shown that gastroenteropancreatic tumors, with the exception of insulinomas, have a high density of somatostatin receptors and can be imaged in vivo using somatostatin receptor scintigraphy (SRS) with either [123I-Tyr3]octreotide or [111In DTPA,DPhe1]octreotide. However, the sensitivity in relation to conventional imaging studies (ultrasound, CT, MRI, angiography) remains unclear. To address this question, we performed a prospective study of 80 patients with gastrinomas where SRS was compared with other conventional imaging techniques for detecting extrahepatic gastrinomas or liver metastases. Extrahepatic gastrinomas were identified by SRS in 58 percent of patients, whereas conventional imaging studies detected gastrinomas in 9 percent to 48 percent of patients. In detecting hepatic metastases in 24 patients with histologically-proven metastases, SRS was positive in 92 percent; ultrasound, CT or angiography in 42 percent to 62 percent; and MRI in 71 percent of patients. These results are compared with other studies in detecting gastrinomas as well as series including other PETs, excluding insulinomas, with insulinomas alone, and with carcinoid tumors. An analysis of the ability of SRS to identify gastrinomas found in different sites at surgery was performed. The role of endoscopic ultrasound (EUS) in detecting various PETs, in comparison to that of SRS, is yet to be established, particularly for extrapancreatic PETs. Therefore, the results of EUS in various studies containing patients with PETs are compared to those with SRS and conventional imaging studies. These data suggest that EUS is the first choice of localization methods for detecting insulinoma, which is an intrapancreatic tumor in almost all cases. In other PETs there still is not sufficient data to establish the relative roles of EUS and SRS.