Objectives: Alzheimer’s disease (AD) is the main cause of gradual cognitive impairment in elderly individuals. This highlights the need of obtaining biomarkers to identify features that are different among mild cognitive impairment (MCI), AD and cognitively normal (CN) individuals. Design and methods: Ultra-performance liquid chromatography (UPLC)/mass spectrometry (MS) was employed to find the metabolic changes in plasma samples obtained from AD, MCI and CN individuals. Based on principal component analysis (PCA), the metabolic differences among AD, MCI and CN subjects were identified. Results: The PCA of UPLC/MS spectra indicated metabolic differences among AD, MCI and CN subjects. The peak intensities of progesterone, lysophos- phatidylcholines (LPCs), tryptophan, L-phenylalanine, dihydrosphingosine and phytosphingosine in the plasma of the MCI and AD subjects were significantly different from the CN subjects. Furthermore, the peak intensities of tryptophan, LPCs, dihydrosphingosine in the plasma of the AD subjects were significantly lower compared to the MCI subjects. Conclusion: Our data provide a link between metabolite imbalance and AD, and suggest that metabolomics can be used to reveal the early disease mechanisms occurred in the progression from CN to MCI and AD.