While mitochondria maintain essential cellular functions, such as energy production, calcium homeostasis, and regulating programmed cellular death, they also play a major role in pathophysiology of many neurological disorders. Furthermore, several neurodegenerative diseases are closely linked with synaptic damage and synaptic mitochondrial dysfunction. Unfortunately, the ability to assess mitochondrial dysfunction and the efficacy of mitochondrial-targeted therapies in experimental models of neurodegenerative disease and CNS injury is limited by current mitochondrial isolation techniques. Density gradient ultracentrifugation (UC) is currently the only technique that can separate synaptic and non-synaptic mitochondrial sub-populations, though small brain regions cannot be assayed due to low mitochondrial yield. To address this limitation, we used fractionated mitochondrial magnetic separation (FMMS), employing magnetic anti-Tom22 antibodies, to develop a novel strategy for isolation of functional synaptic and non-synaptic mitochondria from mouse cortex and hippocampus without the usage of UC. We compared the yield and functionality of mitochondria derived using FMMS to those derived by UC. FMMS produced 3x more synaptic mitochondrial protein yield compared to UC from the same amount of tissue, a mouse hippocampus. FMMS also has increased sensitivity, compared to UC separation, to measure decreased mitochondrial respiration, demonstrated in a paradigm of mild closed head injury. Taken together, FMMS enables improved brain-derived mitochondrial yield for mitochondrial assessments and better detection of mitochondrial impairment in CNS injury and neurodegenerative disease.