NMR-based metabonomics is a valuable and straightforward approach to measuring hundreds of metabolites in complex biofluids. However, metabolite identification is sometimes limited by overlapped signals in NMR spectra. We describe a new methodology using an automated hyphenation of solid phase extraction (SPE) with RP-HPLC combined to NMR spectroscopy, which allowed identification of 72 metabolites of various molecular classes in human urine. This methodology was also Successfully applied to the fractionation of a cat urine sample to aid identification of aromatic Compounds and felinine. The SPE-RP-HPLC method appears to be a reliable tool to support biomarker discovery in metabonomic studies. (C) 2008 Elsevier B.V. All rights reserved.