Drug allergies are a major problem in the clinic and during drug development. At the present time, it is not possible to predict the potential of a new chemical entity to produce an allergic reaction (hypersensitivity) in patients in preclinical development. Such adverse reactions, because of their idiosyncratic nature, only become apparent once the drug has been licensed. Our present chemical understanding of drug hypersensitivity is based on the hapten hypothesis, in which covalent binding of the drug (metabolite) plays a central role in drug immunogenicity and antigenicity. If this theory is correct, then it should be possible to develop in vitro systems to assess the potential of drugs to bind to critical proteins, either directly or indirectly after metabolic activation to protein-reactive metabolites (bioactivation) and initiate hypersensitivity. The purpose of this review is to assess critically the evidence to support the hapten mechanism, and also to consider alternative mechanisms by which drugs cause idiosyncratic toxicity.