Despite extensive preoperative staging, a significant number of pancreatic cancers are unresectable at surgical exploration. Patients undergoing pancreatic exploration with a view to resection were studied and comparisons are then made between those undergoing resection and a bypass procedure to identify surrogate markers of unresectability. One hundred thirteen consecutive patients underwent pancreatic exploration for head-of-pancreas (HOP) adenocarcinoma with curative intent. Fifty-five underwent pancreaticoduodenectomy and 58 underwent a bypass procedure. Student's t test, receiver operator characteristics (ROC) and logistic regression were used to compare the predictive value of preoperative patient variables collected retrospectively. The bypass group had a significantly higher median CA19.9 than the resection group (P = 0.003). Platelet count and neutrophil-lymphocyte ratio (NLR) were also significantly different (P = 0.013 and P = 0.026, respectively). ROC analysis indicated that age < or =65, platelet count >297 x 10(9)/l, CA19.9 < or =473 Ku/l, and CA19.9-bilirubin ratio were predictive variables for resectable disease. NLR and CA19.9-bilirubin ratio had specificity values of 92.9 and 97.0%, respectively. From logistic regression, a raised CA19.9 was found to be an independent risk factor for unresectable disease (P = 0.031). A significant proportion of patients with HOP adenocarcinoma are understaged preoperatively. Preoperative serology including platelet count, NLR, CA19.9, and CA19.9-bilirubin ratio may be used as additional discriminators of resectability particularly for high-risk patients.