Malignant melanoma is one of the most common and aggressive tumors in the oral cavity of dog. The tumor has a poor prognosis, and methods for diagnosis and prediction of prognosis after treatment are required. Here, we examined metabolite profiling using gas chromatography-mass spectrometry (GC-MS) for development of a discriminant model for evaluation of prognosis. Metabolite profiles were evaluated in healthy and melanoma plasma samples using orthogonal projection to latent structure using discriminant analysis (OPLS-DA). Cases that were predicted to be healthy using the OPLS discriminant model had no advanced lesions after radiation therapy. These results indicate that metabolite profiling may be useful in diagnosis and prediction of prognosis of canine malignant melanoma.