Background: KRAS monitoring provides valuable information for early diagnosis and prediction of treatment outcome in colorectal cancer. KRAS mutation is observed in only half of colon cancer patients, whereas it is detected in 90% of pancreatic cancer patients. Therefore, investigating tumor DNA in serum by KRAS monitoring may be even more valuable in pancreatic cancer patients. In this study, we elucidated the clinical significance of KRAS monitoring in pancreatic cancer patients during treatment. Methods:KRAS in tumor tissues was analyzed for mutations by Scorpion ARMS or RASKET in 69 patients with pancreatic tumors. KRAS in serum was analyzed for mutations (G12D, G12V, G12C, G12A, G12S, G12R, G13D, Q61L, and Q61H) using droplet digital polymerase chain reaction in 58 patients who underwent the curative surgery (N = 39) or the chemotherapy (N = 19) and who had KRAS mutation in their tissues. Results:KRAS mutation in tumor tissues was detected in 62 of 69 patients (92.5%). These 62 patients showed significantly poorer prognosis (3years overall survival; 43.9%) than the seven patients without mutation (p = 0.03), who were all alive. Monitoring of KRAS in serum revealed KRAS mutation in 22 of 58 patients (37.9%). In patients who underwent the chemotherapy (N = 19), 2years OS of patients who detected KRAS mutation in serum (N = 9) was 0% and them which not detected it (N = 10) was 46.7% (p = 0.01). And in the curative resection group (N = 39), we detected KRAS mutations in serum among recurrent patients after surgery, but did not detect them among non-recurrent patients. Conclusions:KRAS mutation in serum could be a valuable predictive and prognostic biomarker in pancreatic cancer patients. Additionally, assessment of KRAS in tumor tissues may provide information about individual tumor biology. So, monitoring KRAS status of patients with pancreatic cancer may be useful of the treatment strategy.