Pancreatic cancer is one of the most fatal malignancies in the world with an overall 5-year survival rate of less than 5%. Diagnosing pancreatic cancer is not always straightforward. Clinical imaging of the pancreas can be misleading and currently available clinical biomarkers are few and lack sensitivity and specificity. Obtaining tissue or cytology from the pancreaticobiliary system to confirm a diagnosis is clinically invasive and may be inconclusive. Further non-invasive diagnostic biomarkers are clearly required. I describe the feasibility of nuclear magnetic resonance (NMR) spectroscopy as a modality for novel plasma and urine pancreaticobiliary biomarker discovery Plasma and urine samples from 44 patients undergoing pancreatic resection for pancreaticobiliary malignancy along with a benign cohort of 45 patients were acquired. Spectra were obtained on a Varian NMR 500 MHz spectrometer. Unsupervised and supervised multivariate pattern recognition techniques were used for chemometric analysis. Model validation was assessed through permutation and cross validation techniques Plasma metabonomic profiling identified clear separation between malignant and benign pancreaticobiliary disease with an overall sensitivity and specificity of 64.9 and 73.5% respectively. Sensitivity and specificity among non-jaundiced patients rose to 75 and 75.8% respectively. Suppressed metabolites among cancer patients included VLDL, valine and acetate. Up- regulated metabolites included isobutyrate, 3-hydroxybutyrate, lactate, acetoacetate, pyruvate, glucose and taurine. Urinary metabonomic profiling failed to satisfactorily discriminate between benign and malignant disease. Plasma nuclear magnetic resonance metabonomic profiling has significant potential for future pancreaticobiliary biomarker development. Plasma bilirubin is an important confounding factor, which must be accounted for in all future pancreaticobiliry metabonomic studies.