A computer-based pattern recognition (PR) approach has been applied to the interpretation of 1H NMR generated urinalysis data in a variety of experimental toxicity states in the rat. 1H NMR signal intensities for each endogenous metabolite in urine were regarded as coordinates in multi-dimensional space and analysed using computer pattern recognition methods through which the dimensionality was reduced for display and categorization purposes. Initially 17 metabolic dimensions were used which were defined by the scored relative concentrations of a variety of urinary metabolites detected in 1H NMR spectra. By employing the unsupervised learning methods of 2- and 3-dimensional nonlinear mapping (NLM) different types of toxin (hepatotoxins, cortical and papillary nephrotoxins) could be classified according to NMR-detectable biochemical effects in the urine. The robustness of the classification methods, and the influence of the addition of new scored biochemical data reflecting dose response situations, nutritional effects on toxicity, sex differences in biochemical response to toxins and addition of a new toxin class (testicular toxin) to the pattern recognition analysis were also evaluated. We find that the initial training set maps are fundamentally stable to the addition of all data types and that the PR methods correctly 'predicted' the toxicological effects of the test compounds. These results confirm the power and wide applicability of linked PR and 1H NMR urinalysis as an approach to the generation and classification of acute toxicological data.