The diagnostic utility of fluorine-18 2-deoxy-D-glucose positron emission tomography (FDG PET) for the non-invasive differentiation of focal pancreatic lesions originating from cancer or chronic pancreatitis by combined visual image interpretation and semiquantitative uptake value analysis has been documented. However, in clinical routine some misdiagnosis is still observed. This is because there is potential overlap between the semiquantitative uptake values obtained for active inflammatory lesions and cancer. Therefore, this prospective study was undertaken to test the hypothesis that analysis of dynamic kinetics of focal pancreatic lesions based on FDG PET may more accurately determine the benign or malignant nature of such lesions. Thirty patients (56+/-17 years) were studied dynamically with FDG PET for a period of 60-90 min. Patients were assigned to one of four groups: control, acute pancreatitis, chronic pancreatitis or pancreatic cancer. Two observers, blinded to the clinical data, analysed the time-activity curves of FDG kinetics based on region of interest analysis. The diagnosis predicted by FDG PET was compared with the result of histological examination of the surgical specimen. Analysis of FDG kinetics revealed significant differences in the shape of the time-activity curve for controls, pancreatic cancer and inflammatory disease. Surprisingly, there was no significant difference in the time-activity curve shape for chronic pancreatitis and acute pancreatitis; this is, however, not a clinical issue. Furthermore, acquisition time (60 min vs 90 min) did not affect interpretation of the time-activity curve, so that scanning time may be regularly shortened to 60 min. Interobserver agreement was 1. Based on these findings, non-invasive differentiation between pancreatic cancer and chronic pancreatitis was correctly predicted in all cases, as confirmed by histology. In addition, the specificity was increased compared with that obtained from standardised uptake value analysis. Non-invasive differentiation between pancreatic cancer and chronic pancreatitis may best be achieved based on a dynamic FDG PET study including kinetic analysis. This approach yields results superior to those obtained from a semiquantitative analysis of pancreatic lesions.