Activation by diazoxide and inhibition by 5-hydroxydecanoate are the hallmarks of mitochondrial ATP-sensitive K+ (K(ATP)) channels. Opening of these channels is thought to trigger cytoprotection (preconditioning) through the generation of reactive oxygen species. However, we found that diazoxide-induced oxidation of the widely used reactive oxygen species indicator 2',7'-dichlorodihydrofluorescein in isolated liver and heart mitochondria was observed in the absence of ATP or K+ and therefore independent of K(ATP) channels. The response was blocked by stigmatellin, implying a role for the cytochrome bc1 complex (complex III). Diazoxide, though, did not increase hydrogen peroxide (H2O2) production (quantitatively measured with Amplex Red) in intact mitochondria, submitochondrial particles, or purified cytochrome bc1 complex. We confirmed that diazoxide inhibited succinate oxidation, but it also weakly stimulated state 4 respiration even in K+-free buffer, excluding a role for K(ATP) channels. Furthermore, we have shown previously that 5-hydroxydecanoate is partially metabolized, and we hypothesized that fatty acid metabolism may explain the ability of this putative mitochondrial K(ATP) channel blocker to inhibit diazoxide-induced flavoprotein fluorescence, commonly used as an assay of K(ATP) channel activity. Indeed, consistent with our hypothesis, we found that decanoate inhibited diazoxide-induced flavoprotein oxidation. Taken together, our data question the "mitochondrial K(ATP) channel" hypothesis of preconditioning. Diazoxide did not evoke superoxide (which dismutates to H2O2) from the respiratory chain by a direct mechanism, and the stimulatory effects of this compound on mitochondrial respiration and 2',7'-dichlorodihydrofluorescein oxidation were not due to the opening of K(ATP) channels.