Although it is widely believed that generation of reactive oxygen species (ROS) makes an important contribution to ischaemic injury in the brain, the mechanisms and time course of ROS generation during ischaemia and reperfusion have not been clearly defined. We have used fluorescence imaging techniques to examine the generation of ROS during oxygen glucose deprivation (OGD). Rates of ROS generation were measured using dihydroethidine (Het) in mixed cultures of glia and neurons from rat hippocampus or cortex. Statistical analysis and exponential curve fitting were performed using Origin 7 software. Results are expressed as means ± standard error of the mean (S.E.M.). In response to OGD with oxygen levels sufficient to cause loss of mitochondrial potential, we detected an early increase in the rate of ROS production occurring after 3-7 min in the majority of neurons (an increase in rate of 1.58-fold; p<0.05; n=215). This increase was followed by a prolonged period (lasting 20-25 min) during which rates of ROS generation were reduced to 36±4.3 of the basal rate in cortical cells. This was then followed by a secondary increase in ROS generation starting after 23-35 min of OGD increasing to 2.1-fold of the basal rate of cortical neurons. Reoxygenation increased the rate of ROS production yet again, to 2.56-fold basal rate in hippocampal and 2.89-fold in cortical neurons. Each of these phases of ROS production seems to reflect a different mechanism as each was inhibited by different compounds: the first phase by the protonophore FCCP (1μM, n=155), the delayed secondary phase by the inhibitor of xanthine oxidase, oxypurinol (20 μM, n=165), while the increase in ROS generated upon reoxygenation was blocked by the inhibitor of the NADPH oxidase, DPI (0.5 μM, n=201), and was absent in cells cultured from gp91phox knockout mice. The response to mitochondrial inhibition by 1mM NaCN with 2μg/ml oligomycin was qualitatively similar but quantitatively quite different. Thus, over 10 min after CN/oligomycin application, the rate of ROS generation increased dramatically (3.35-fold in cortical neurons, p<0.001; n=78). This was again followed after 10-20 min by a reduction of the rate of ROS production below basal levels (40.4±4.6% of basal rate in hippocampal neurons) to be followed later by a secondary activation of ROS production (3.14-fold, p<0.001, in cortical neurons). These data strongly suggest that ROS is generated by three quite distinct processes at different times during OGD and also show that ROS generation by metabolic inhibition in the presence of oxygen (CN) is quite different to ROS generation by OGD. We suggest that early ROS production is derived from the mitochondrial respiratory chain. Reduction in ROS generation then correlates with collapse of the mitochondrial potential. Once ATP is depleted, generating AMP and hypoxanthine, xanthine oxidase becomes active generating more ROS despite continued hypoxia. ROS generation at reperfusion is primarily due to activation of the NADPH oxidase.