Objective To investigate the clinical value of a preoperative predictive scoring system established by National Cancer Center Hospital (NCCH) for postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD).Methods The clinical data of 100 patients with periampullary tumor who underwent PD in the Cancer Hospital of Tianjin Medical University from October 2008 to January 2012 were retrospectively analyzed.Five indexes including gender,pancreatic cancer,main pancreatic duct index,portal invasion and intra-abdominal fat thickness were in the NCCH preoperative predictive scoring system.Patients with score ＞ 4 were defined as with high risk of POPF,and those with score≤4 were defined as with low risk of POPF.Factors associated with POPF were analyzed using the Pearson chi-square test.The sensitivity and specificity of the predictive scoring system were determined by receiver operating characteristic (ROC) curve analysis.Results Of the 100 patients,20 had POPF,including 9 in grade A,6 in grade B and 5 in grade C.Gender,pancreatic cancer,portal invasion,texture of pancreas and method of pancreaticojejunostomy were closely correlated with POPF (x2=5.613,4.785,15.479,7.145,7.050,P ＜ 0.05).The incidence of POPF was 86.4％ (19/22) for patients with high risk of POPF,and 1.3％ (1/78) for patients with low risk of POPF,with significant difference (x2=77.637,P ＜ 0.05).The results of ROC curve analysis showed that the sensitivity and specificity of the predictive scoring system were 95.0％ and 96.3％,respectively.The nomogram showed an area under the curve of 99.0％ (P ＜ 0.05).Conclusion The NCCH preoperative predictive scoring system could accurately predict the occurrence of POPF.