Several lines of evidence indicate that mitochondrial reactive oxygen species (ROS) generation is the major source of oxidative stress in the cell. It has been shown that ROS production accompanies cytochrome c release in different apoptotic paradigms, but the site(s) of ROS production remain obscure. In the current study, we demonstrate that loss of cytochrome c by mitochondria oxidizing NAD(+)-linked substrates results in a dramatic increase of ROS production and respiratory inhibition. This increased ROS production can be mimicked by rotenone, a complex I inhibitor, as well as other chemical inhibitors of electron flow that act further downstream in the electron transport chain. The effects of cytochrome c depletion from mitoplasts on ROS production and respiration are reversible upon addition of exogenous cytochrome c. Thus in these models of mitochondrial injury, a primary site of ROS generation in both brain and heart mitochondria is proximal to the rotenone inhibitory site, rather than in complex III. ROS production at complex I is critically dependent upon a highly reduced state of the mitochondrial NAD(P)(+) pool and is achieved upon nearly complete inhibition of the respiratory chain. Redox clamp experiments using the acetoacetate/L-beta-hydroxybutyrate couple in the presence of a maximally inhibitory rotenone concentration suggest that the site is approx. 50 mV more electronegative than the NADH/NAD(+) couple. In the absence of inhibitors, this highly reduced state of mitochondria can be induced by reverse electron flow from succinate to NAD(+), accounting for profound ROS production in the presence of succinate. These results lead us to propose a model of thermodynamic control of mitochondrial ROS production which suggests that the ROS-generating site of complex I is the Fe-S centre N-1a.