The occurrence of drug-induced liver injury (DILI) presents a significant safety issue for patients and represents a major cause of regulatory action. The methods that are in current use for early detection and prediction of DILI in patients are not adequate. The liver is the major site of synthesis of endogenous metabolites, and data suggest that alterations in the profiles of endogenous metabolites ("the metabolome") may precede development of clinically overt DILI. Metabonomics involves the application of analytical technologies such as nuclear magnetic resonance and mass spectrometry to detect changes in the metabolome. In this review, we describe the emerging role of metabonomics in predicting and understanding the mechanisms underlying DILI. Recent human clinical trials of drugs, including acetaminophen (APAP) and ximelagatran, have shown that the metabonomics of biofluids (plasma and urine) collected before and immediately after dosing can identify individual patients who are likely to develop DILI. These studies support the need to include metabonomic investigations in clinical trials of potentially hepatotoxic medications.