Several etiologies result in chronic liver diseases including chronic hepatitis C virus infection (HCV). Despite its high incidence and the severe economic and medical consequences, liver disease is still commonly overlooked due to the lack of efficient non-invasive diagnostic methods. While several techniques have been tested for the detection of fibrosis, the available biomarkers still present severe limitations that preclude their use in clinical diagnostics. Liver diseases have also been the subject of metabolomic analysis. Here, we demonstrate the suitability of .sup.1 H NMR spectroscopy for characterizing the metabolism of liver fibrosis induced by HCV. Serum samples from HCV patients without fibrosis or with liver cirrhosis were analyzed by NMR spectroscopy and the results were submitted to multivariate and univariate statistical analysis. PLS-DA test was able to discriminate between advanced fibrotic and non-fibrotic patients and several metabolites were found to be up or downregulated in patients with cirrhosis. The suitability of the most significantly regulated metabolites was validated by ROC analysis. Our study reveals that choline, acetoacetate and low-density lipoproteins are the most informative biomarkers for predicting cirrhosis in HCV patients. Our results demonstrate that statistical analysis of .sup.1 H-NMR spectra is able to distinguish between fibrotic and non-fibrotic patients suffering from HCV, representing a novel diagnostic application for NMR spectroscopy.