This study employed proteomic and bioinformatic approaches to identify serum biomarkers in canine lymphoma patients. Chilled serum samples derived from non-lymphoma ( n = 92) and lymphoma ( n = 87) patients were shipped from first opinion veterinary practices, subjected to ion exchange chromatography and analysed by surface-enhanced laser desorption ionization mass spectrometry. Nineteen serum protein peaks were identified between the two groups as being significantly different ( P < 0.05) based upon their normalized ion intensities. Two biomarkers were identified that were capable of differentiating lymphoma and non-lymphoma patients. Analysis of the test data provided a positive predictive value (PPV) of 82%. A clinical follow-up study was carried out on 96 canine patients suspected of having lymphoma. Evaluation of this data gave a specificity value of 91%, sensitivity of 75%, PPV of 80% and negative predictive value of 88%. In conclusion, the expression pattern of two serum biomarkers has enabled serum samples to be classified into either lymphoma or non-lymphoma categories. [ABSTRACT FROM AUTHOR]