KRAS2 mutations in codon 12 have been detected in about 80% of pancreatic cancers. The aim of this study was to evaluate the value of KRAS2 mutations detection in circulating deoxyribo nucleic acid to differentiate pancreatic cancer from chronic pancreatitis. Circulating deoxyribonucleic acid was isolated from serum in 47 patients with histologically proven pancreatic adenocarcinomas (26 males, median age 65 years) and 31 controls with chronic pancreatitis (26 males, median age 48 years). Mutations at codon 12 of KRAS2 gene were searched for using polymerase chain reaction and allele specific amplification. Serum carbohydrate antigen 19.9 levels were also determined. KRAS2 mutations were found in 22 patients (47%) with pancreatic cancer and in four controls with chronic pancreatitis (13%) (P<0.002). None of the latter developed a pancreatic cancer within the 36 months of median follow-up. The sensitivity, specificity, positive and negative predictive values of serum serum KRAS2 mutations for the diagnosis of pancreatic cancer were 47, 87, 85 and 52%, respectively. KRAS2 mutations were not related to age, gender, smoking habit, tumour stage, or survival. Among the 26 patients with normal or non-contributive (due to cholestasis) serum carbohydrate antigen 19.9 levels, 14 (54%) had KRAS2 mutations. The combination of KRAS2 and carbohydrate antigen 19.9 gave a sensitivity, specificity, positive and negative predictive values for the diagnosis of pancreatic cancer of 98, 77, 87 and 96%, respectively. Detection of KRAS2 mutations in circulating deoxyribo nucleic acid has a low sensitivity but a specificity about 90% for the diagnosis of pancreatic cancer. It seems particularly useful when serum carbohydrate antigen 19.9 levels are normal or inconclusive. A combined normal serum carbohydrate antigen 19.9 and absence of circulating KRAS2 mutations makes the diagnosis of pancreatic cancer extremely unlikely.