Adverse drug reactions, such as hepatotoxicity, blood dyscrasias and hypersensitivity are a major obstacle for the use and the development of new medicines. Many forms of organ-directed toxicity can arise from the bioactivation of drugs to the so-called chemically reactive metabolites, which can modify tissue macromolecules. It is well established that the toxicities of model hepatotoxins, such as acetaminophen, furosemide, bromobenzene and methapyrilene can be correlated with the generation of chemically reactive metabolites, which can be detected by measurement of the irreversible binding of radiolabelled material to hepatic protein and/or the detection of stable phase II metabolites such as glutathione conjugates. The basic chemistry of the reaction of such metabolites with model nucleophiles is relatively well understood. A major challenge is to define how certain reactive intermediates may chemically modify critical proteins and how modification of specific amino acids may alter protein function which in turn may affect cell signalling, regulation, defence, function and viability. This in turn will determine whether or not bioactivation will result in a particular form of drug-induced injury. It is now clear that even relatively simple reactive intermediates can react in a discriminative manner with particular cellular proteins and even with specific amino acids within those proteins. Therefore, both non-covalent, as well as covalent bonds will be important determinants of the target protein for a particular reactive metabolite. Mammalian cells have evolved numerous defence systems against reactive intermediates. Sensitive redox proteins such as Nrf-2 recognise oxidative stress and electrophilic agents, through oxidation or covalent modification of thiol groups. Defence genes, such as epoxide hydrolase and glutamate cysteine ligase then become up-regulated in an attempt to reduce the oxidising environment. However, whether the liver receives mild or severe injury depends upon extra-cellular signalling processes between the hepatocytes and non-parenchymal cells, particularly kupffer and natural killer cells (NK/NK T cells). Determination of the nature and downstream effect of these extra-cellular signalling processes is critical in order to design better predictive hepatotoxicity screens. More importantly, to understand and manipulate these signalling processes will aid in the design of safer therapeutic agents, but also contribute to the clinical management of liver disease.