Non-alcoholic steatohepatitis (NASH) is a common liver disease. NASH is characterized by fatty liver, along with inflammation. Most people with NASH are not aware of their condition, even though NASH can lead to hepatic cirrhosis. Several approaches have been tested to clarify the pathology of NASH. However, the mechanism of onset of NASH was not well-defined. In this study, a supervised multivariate analysis (principal component analysis) approach using direct matrix-assisted laser desorption/ionization mass spectrometry (dMALDI-MS) was applied to the analysis of metabolites in starvation-induced fatty liver tissue sections. This approach does not require complex pretreatments. We investigated the characteristic dynamics of metabolites in fatty liver. This approach can be applied to the analysis of human biopsy specimens of fatty liver in future studies.