Identification of pancreatic cancer in patients presenting with an enlarged pancreatic mass is a major diagnostic problem. Positron emission tomography (PET) using the radiolabeled glucose analogue 18F-fluorodeoxyglucose (FDG) has been suggested to provide excellent accuracy for noninvasive determination of suspicious pancreatic masses. We conducted a prospective study to verify these results. Forty-two patients admitted for pancreatic surgery underwent PET scanning. Image analysis was based on visual film evaluation and quantification of regional tracer uptake. PET imaging was visually analyzed by three observers blinded for the results of other diagnostic tests; they qualitatively graded the scans using a five-point scale (I = low to V = high) for the presence and intensity of focal FDG uptake. Diagnosis was proven by histology (n = 38) or follow-up (n = 4). Furthermore, the results of PET were compared with helical computed tomography (CT) and conventional ultrasonography (US), done during the routine diagnostic workup before pancreatic cancer surgery. Regarding only the results with scores of IV and V as positive for representing definite malignancy yielded a sensitivity of 71% and a specificity of 64% for film reading. Quantification of regional tracer uptake contributed no significant diagnostic advantage for differentiation between benign and malignant tumors. Helical CT revealed a sensitivity of 74% and a specificity of 45.5% and abdominal US 56% and 50%, respectively. We concluded that PET imaging provides only fair diagnostic accuracy (69%) for characterizing enlarged pancreatic masses. PET does not allow exclusion of malignant tumors. In doubtful cases, the method must be combined with other imaging modalities, such as helical CT. The results indicate that the number of invasive procedures is not significantly reduced by PET imaging.