Imaging of cerebral oxygen metabolism is useful for selecting and monitoring therapy, matching patients with interventions, and optimizing outcomes while reducing the incidence of hemorrhagic changes. We aimed to measure oxygen metabolism using magnetic resonance (MR) imaging. Because the magnetic susceptibility of hemoglobin varies with its redox status, the regional oxygen saturation should be measurable by MR imaging. We derived equations based on the blood oxygenation level-dependent (BOLD) theory and evaluated their validity in patients with chronic cerebral ischemia. From changes in BOLD signal, cerebral blood flow (CBF), and cerebral blood volume (CBV) after loading of acetazolamide or breath-holding, we calculated the cerebral metabolic rate of oxygen (CMRO(2)) pixel by pixel and compared the results to those of positron emission tomography (PET) examination. The correlation coefficient between the 2 modalities was between 0.70 and 0.75 in 4 patients, suggesting the applicability of BOLD-contrast MR imaging, but ranged from 0.39 to 0.47 in assessment of oxygen extraction fraction (OEF). BOLD contrast MR imaging tended to overestimate the region of decreased OEF. Some technical challenges remain for improving the sensitivity of delta BOLD, selective measurement of venous blood volume, and shortening of examination time.