Mitochondrial complex I has previously been shown to release superoxide exclusively towards the mitochondrial matrix, whereas complex III releases superoxide to both the matrix and the cytosol. Superoxide produced at complex III has been shown to exit the mitochondria through voltage dependent anion channels (VDAC). To test whether complex I-derived, mitochondrial matrix-directed superoxide can be released to the cytosol, we measured superoxide generation in mitochondria isolated from wild type and from mice genetically altered to be deficient in MnSOD activity (TnIFastCreSod2(fl/fl)). Under experimental conditions that produce superoxide primarily by complex I (glutamate/malate plus rotenone, GM+R), MnSOD-deficient mitochondria release ∼4-fold more superoxide than mitochondria isolated from wild type mice. Exogenous CuZnSOD completely abolished the EPR-derived GM+R signal in mitochondria isolated from both genotypes, evidence that confirms mitochondrial superoxide release. Addition of the VDAC inhibitor DIDS significantly reduced mitochondrial superoxide release (∼75%) in mitochondria from either genotype respiring on GM+R. Conversely, inhibition of potential inner membrane sites of superoxide exit, including the matrix face of the mitochondrial permeability transition pore and the inner membrane anion channel did not reduce mitochondrial superoxide release in the presence of GM+R in mitochondria isolated from either genotype. These data support the concept that complex I-derived mitochondrial superoxide release does indeed occur and that the majority of this release occurs through VDACs.