Although development of insulin therapy prolong patients` lifespan, there are still diabetic complications which are not reverted by insulin. Many diabetic complications are caused by accumulation of sorbitol which is generated by aldose reductase. Hence, many aldose reductase inhibitor are developed as therapeutic agents of diabetic complications. However, there is no globally approved aldose reductase due to severe adverse effects like hepatotoxicity. Even molecular mechanism of adverse effects of aldose reductase inhibitors are not investigated. This research aimed to investigate molecular mechanisms of aldose reductase inhibitors` adverse effect using QSAR. QSAR is computational approaches to predict biological or chemical properties from characteristics which can be measured easily. QSAR models of various enzymes which are known as hepatotoxicity-related targets are built from ChEMBL database. Then inhibitory activity of aldose reductase inhibitors are predicted with these QSAR models. Finally, off-target candidates which are significantly well inhibited by aldose reductase is selected as putative off-targets of aldose reductase. As a result, cyclooxygenase 1 which is one of key enzyme of cellular signal pathway is chosen as putative off-targets of aldose reductase inhibitors. inhibitors.