Troglitazone has been withdrawn from therapeutic options for diabetes mellitus because of its severe hepatocyte toxicity of unknown pathogenesis. The aim of the present study was to assess both morphologic and functional alterations in the mitochondria of troglitazonetreated hepatocytes. A polarized human hepatocyte cell line, OUMS-29, was used in this study. The mitochondrial volume and the mitochondrial transmembrane potential (⌬⌿ m ) were examined using flow cytometry with nonylacridine orange (NAO) and rhodamine-123, respectively. An ultrastructural examination of the troglitazone-treated OUMS-29 cells was performed using transmission electron microscopy (TEM). Reactive oxygen species (ROS) production was assessed using flow cytometry with dihydroethidium and 2 ,7 -dichlorodihydrofluorescein diacetate. A significant increase in the mitochondrial volume of the troglitazone-treated cells was found by the NAO analysis, in comparison with pioglitazone-treated and ciglitazone-treated cells. The increase in volume was due to a marked enlargement in the mitochondria. The markedly enlarged mitochondria with intramitochondrial electrondense deposits were confirmed on TEM, which showed myelin-like structures, indicating degraded membrane constituents. The troglitazone-treated cells showed a significant decline in the ⌬⌿ m per unit mitochondrial volume but resulted in no clear cell death. ROS analysis revealed a significant production of hydrogen peroxide in the troglitazone-treated hepatocytes. This production was attenuated using an antioxidant, N-acetyl-L-cysteine. In conclusion, troglitazone caused overproduction of hydrogen peroxide, which deteriorated both mitochondrial membrane structures and mitochondrial function, leading to a possible priming for the severe hepatocyte toxicity. (HEPATOLOGY 2003;37:136-147.) T roglitazone, a ligand for peroxisome proliferatoractivated receptor ␥ (PPAR␥), used to be administered as an antidiabetic drug until it was discovered to be associated with hepatotoxicity and withdrawn from the market after reports of several dozens of deaths or cases of severe hepatic failure requiring liver transplantation. 1,2 In contrast to troglitazone, 2 other glitazones, pioglitazone and rosiglitazone, are still on the market; however, several cautionary reports have suggested that the latter 2 drugs show hepatotoxicity as a class effect of glitazones. [2][3][4][5][6][7][8][9] Although the troglitazone-associated hepatotoxicity has been considered idiosyncratic direct hepatocyte toxicity, 10 the precise mechanism has remained unclear. In our recent study, mitochondrial damage was ultrastructurally demonstrated in troglitazone-treated human hepatoma cells, 11 allowing us to first focus on alterations in the mitochondrial structure and the function in the troglitazone-treated human hepatocytes. Troglitazone is known