The rapid evolution of drug discovery science, fuelled by combinatorial librarybased synthesis programmes, has led to increased pressure on the drug safety evaluation process. Once potential drugs have passed the primary biological screening procedures, losses of drug candidate compounds from the product development pipeline (known asattrition’) need to be minimized. Hence, there is an intensive search for new analytical technologies that will maximize eæciency of lead compound selection based both on eæcacy and safety and will minimize overall attrition rates. Current bioanalytical approaches include measurements of responses of living systems to drugs either at the genetic level or at the level of expression of cellular proteins, using so-called genomic and proteomic methods respectively. At present both genomics and proteomics are expensive and labour-intensive, yet potentially are powerful tools for studying diåerent levels of the biological response to xenobiotic exposure. However, even in combination, genomics and proteomics do not provide the range of information needed for an understanding of the integrated cellular function in living systems, since both ignore the dynamic metabolic status of the whole organism. Thus, a new NMR-basedmetabonomic’approach is proposed that is aimed at the augmentation and complementation of the information provided by measuring the genetic and proteomic responses to xenobiotic exposure. Metabonomics is deŪned asthe quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modiŪcation’. This concept has arisen from work on the application of" H-NMR spectroscopy to study the multicomponent metabolic composition of bioŊuids, cells and tissues over the past two decades (eg Nicholson et al. 1983, 1985, Bales et al. 1984, Gartland et al. 1989, Nicholson and Wilson 1989, Moka et al. 1998). Also studies utilizing pattern recognition (PR), expert systems and related bio-informatic tools are used to interpret and classify complex NMR-generated metabolic data sets (Gartland et al. 1991, Holmes et al. 1992, 1994, 1998a, b, Anthony et al. 1994, Spraul et al. 1997, Beckwith-Hall et al. 1998). There is also a signiŪcant background to this work in other research Ūelds, notably metabolic control analysis (Kacser and Burns 1973, Kacser 1993, Goodacre et al. 1996), and there is a related concept of theMetabolome’that represents the total small molecule complement of a cell. However, metabonomics deals with detecting,