Potato (Solanum tuberosum L.) tubers are usually harvested once a year; thus, long-term storage is required to supply quality-assured tubers throughout the year. Further, an applicable method to predict tuber quality during storage is needed. In this study, gas chromatography-mass spectrometry (GC/MS) metabolomics was applied to identify applicable biomarkers for prediction of potato chip color based on 3 years' field-grown tubers. The projections to latent structures (PLS) prediction model, calculated from a metabolome data set obtained before storage, was consistent with actual measured chip color values. Additionally, GC with frame ionization detector (GC/FID) metabolite fingerprinting simultaneously re-constructed more reliable and relevant prediction models for chip color quality compared to GC/MS. Moreover, nine metabolites detected by GC/MS analysis were further validated as applicable prediction markers. This strategy will provide a practical and cost-effective quality-control tool for potato processing manufacturers on an industrial scale.