Devil facial tumor disease (DFTD) is a transmissible cancer threatening Tasmanian devils (Sarcophilus harrisii) with extinction. There is no preclinical test available for DFTD, and thus our aim was to find biomarkers for DFTD by metabolic fingerprinting. Seventy serum samples from wild Tasmanian devils (35 controls, 35 with tumors) were analyzed by liquid chromatography-high-resolution mass spectrometry. Features were selected by multivariate models (PLS/DA, random forests) comparing age-matched training set (n = 20 × 2) and further complying with fold-change threshold (≥1.4) and Mann-Whitney U-tests with correction for multiple hypotheses (false discovery rate (FDR) q < 0.05). An array of overlapping peptide segments of the N-terminal end of fibrinogen were the strongest positive DFTD markers. These peptides recorded fold-change up to 90, FDR-corrected p value below 0.01, and area under ROC curve of at least 0.80 and also correlated with tumor size (Spearman R > 0.45, p < 0.01). Additional potential markers included amino acid and lipid metabolites, while cortisol and urea were the most significant health predictors (AUC ≥ 0.90). PLS/DA resulted in AUC = 0.997 for the training set and overall sensitivity of 91% and specificity of 97%. A support vector machine model utilizing only the major peptide marker and seven other metabolites led to overall 94% sensitivity and specificity. The novel findings in this first DFTD metabolomics study shed light on metabolic changes in Tasmanian devils affected by DFTD and provide a valuable step toward the development of prognostic biomarkers.