Advancements in clinical therapies have identified the need for biomarkers of early Alzheimer disease that distinguish the earliest stages of pathology and target those patients who are likely to gain the most benefit. The aim of this study was to characterize the longitudinal metabolic changes measured by 1H magnetic resonance spectroscopy in correlation to neuropsychologic indices of episodic memory, attention and mental processing speed, language facility, and executive function in subjects with mild cognitive impairment (MCI). Quantitative 1H magnetic resonance spectroscopy of the posterior cingulate gyrus was performed and repeated at 11.56+/-4.3 months. N-acetyl aspartate (NAA), total choline (Cho), total creatine (Cr), myo-inositol (mI), and glutamate/glutamine (Glx) metabolite levels were measured, corrected for cerebrospinal fluid dilution, and ratios calculated in MCI and cognitively normal subjects. In the first study, MCI subjects showed lower NAA levels, NAA/Cho, and NAA/mI ratios and increased Cho/Cr and mI/Cr compared with controls. In the follow-up study, 36% of the MCI subjects [atypical MCI (atMCI)] showed interval increases in NAA, Cr, and Glx levels compared with 64% of MCI subjects (typical MCI) who showed an interval decrease in NAA, Cr, and Glx. Both MCI subgroups had higher Clinical Dementia Rating scores and lower scores on episodic memory, phonemic, and semantic word fluency tasks, compared with controls. The annualized rate of change in metabolic and cognitive status did not differ between normal aging and MCI subjects. atMCI subjects showed significant negative correlations between metabolite levels and executive function task scores, with NAA/mI showing a significant positive correlation with phonemic and semantic word fluency. There were no significant correlations between metabolite levels and cognitive performance in tMCI subjects; however, NAA/mI and mI/Cr were negatively correlated with executive function tasks. These results indicate 2 distinct evolving metabolite profiles that correlate with changes in executive function and can be used to differentiate MCI from normal aging.