The aim of the present prospective observational study was to evaluate the accuracy of preoperative imaging studies and the outcome of patients after pancreaticoduodenectomy for suspected but unproved malignancy. Pancreaticoduodenectomy was performed in 186 patients with a suspected but histologically unproved malignancy of the pancreatic head: 86 with a ductal pancreatic carcinoma, 56 with a periampullary tumor, 18 with a cystadenocarcinoma, 13 with a rare malignant tumor or a metastasis, and 13 with a benign tumor. An accurate differentiation between a ductal pancreatic carcinoma and a nonductal tumor or a benign tumor was neither possible with tumor marker CA 19-9 nor with diagnostic imaging. After pancreatoduodenectomy the postoperative morbidity and mortality were 27.3% and 2.6%, respectively. The median survival time for patients with a ductal pancreatic carcinoma was 9.2 months; for those with nonductal tumors it varied (depending on the histology) between 24.5 and 44.6 months. These results show that a nihilistic approach in the case of tumors of the pancreatic head with suspected but unproved malignancy is not justified. Pancreaticoduodenectomy should be performed for any pancreatic tumor even without histologic confirmation if an experienced pancreatic surgeon cannot exclude pancreatic carcinoma.