We re-examined the reported increase in mitochondrial ROS production during acute hypoxia in cells. Using the Amplex Ultrared/horseradish peroxidase assay we found a decrease, not increase, in hydrogen peroxide release from HEK293 cells under acute hypoxia, at times ranging from 1 min to 3 h. The rates of superoxide/hydrogen peroxide production from each of the three major sites (site IQ in complex I and site IIIQo in complex III in mitochondria, and NADH oxidases (NOX) in the cytosol) were decreased to the same extent by acute hypoxia, with no change in the cells' ability to degrade added hydrogen peroxide. A similar decrease in ROS production under acute hypoxia was found using the diacetyldichlorofluorescein assay. Using a HIF1α reporter cell line we confirmed earlier observations that suppression of superoxide production by site IIIQo decreases HIF1α expression, and found similar effects of suppressing site IQ or NOX. We conclude that increased mitochondrial ROS do not drive the response of HIF1α to acute hypoxia, but suggest that cytosolic H2O2 derived from site IQ, site IIIQo and NOX in cells is necessary to permit HIF1α stabilization by other signals.