Pancreatic cancer is one of the leading causes of cancer-related death, and there is currently little hope of a cure because there are no effective biomarkers for its early detection. Therefore, the search for novel biomarkers that would allow the early detection of pancreatic cancer is ongoing. In this study, the differences between the metabolomes of pancreatic cancer patients with Stage III, Stage IVa, or Stage IVb disease (n = 20) and healthy volunteers (n = 9) were evaluated by metabolomics, which is the endpoint of the Omics cascade and therefore the last step in the cascade before the phenotype. In our experimental conditions using gas chromatography mass spectrometry (GC/MS), a total of 60 metabolites were detected in serum, and the levels of 18 of the 60 metabolites were significantly changed in pancreatic cancer patients compared with those in healthy volunteers. Then, Principal Component Analysis (PCA), which is a basic form of Multiple Classification Analysis, was performed, and the PCA scores plots based on the 60 metabolites highlighted the metabolomic differences between the pancreatic cancer patients and healthy volunteers. The differences between different stages of pancreatic cancer were also assessed by Partial Least Squares Discriminant Analysis (PLS-DA), which is one of Multiple Classification Analysis, and we found that it was possible to discriminate among the Stage III, Stage IVa, and Stage IVb groups. In addition, values of the 9 metabolites in 1 Stage I pancreatic cancer patient were similar to those obtained from the Stage III, Stage IVa, and Stage IVb pancreatic cancer patients. Our findings will aid the discovery of novel biomarkers that allow the early detection of pancreatic cancer by metabolomic approaches.