In a typical metabolomics experiment, two or more conditions (e.g., treated versus untreated) are compared, in order to investigate the potential differences in the metabolic profiles. When dealing with complex biological systems, a two-class classification is often unsuitable, since it does not consider the unpredictable differences between samples (e.g., nonresponder to treatment). An approach based on statistical process control (SPC), which is able to monitor the response to a treatment or the development of a pathological condition, is proposed here. Such an approach has been applied to an experimental hepatocarcinogenesis model to discover early individual metabolic variations associated with a different response to the treatment. Liver study was performed by nuclear magnetic resonance (NMR) spectroscopy, followed by multivariate statistical analysis. By this approach, we were able to (1) identify which treated samples have a significantly different metabolic profile, compared to the control (in fact, as confirmed by immunohistochemistry, the method correctly classified 7 responders and 3 nonresponders among the 10 treated animals); (2) recognize, for each individual sample, the metabolites that are out of control (e.g., glutathione, acetate, betaine, and phosphocholine). The first point could be used for classification purposes, and the second point could be used for a better understanding of the mechanisms underlying the early phase of carcinogenesis. The statistical control approach can be used for diagnosis (e.g., healthy versus pathological, responder versus nonresponder) and for generation of an individual metabolic profile, leading to a better understanding of the individual pathological processes and to a personalized diagnosis and therapy.