Epidemiological studies have identified a robust association between type II diabetes mellitus and Alzheimer disease (AD), and neurobiological studies have suggested the presence of central nervous system insulin resistance in individuals with AD. Given this association, we hypothesized that the central nervous system–penetrant insulin-sensitizing medication metformin would be beneficial as a disease-modifying and/or symptomatic therapy for AD, and conducted a placebo-controlled crossover study of its effects on cerebrospinal fluid (CSF), neuroimaging, and cognitive biomarkers. Twenty nondiabetic subjects with mild cognitive impairment or mild dementia due to AD were randomized to receive metformin then placebo for 8 weeks each or vice versa. CSF and neuroimaging (Arterial Spin Label MRI) data were collected for biomarker analyses, and cognitive testing was performed. Metformin was found to be safe, well-tolerated, and measureable in CSF at an average steady-state concentration of 95.6 ng/mL. Metformin was associated with improved executive functioning, and trends suggested improvement in learning/memory and attention. No significant changes in cerebral blood flow were observed, though post hoc completer analyses suggested an increase in orbitofrontal cerebral blood flow with metformin exposure. Further study of these findings is warranted.