We report a novel two-dimensional (2D) separation strategy aimed at improving the detectability of liquid chromatography mass spectrometry (LC-MS) for metabolome analysis. It is based on the use of ion-pairing (IP) reversed-phase (RP) LC as the first dimension separation to fractionate the metabolites, followed by isotope labeling of individual fractions using dansylation chemistry to alter the physiochemical properties of the metabolites. The labeled metabolites having different hydrophobicity from their unlabeled counterparts are then separated and analyzed by on-line RPLC Fourier-transform ion-cyclotron resonance mass spectrometry (FTICR-MS). This off-line 2D-LC-MS strategy offers significant improvement over the one-dimensional (1D) RPLC MS technique in terms of the number of detectable metabolites. As an example, in the analysis of a human urine sample, 3564 ¹³C-/¹²C-dansylated ion pairs or metabolites were detected from seven IP RPLC fractions, compared to 1218 metabolites found in 1D-RPLC-MS. Using a library of 220 amine- and phenol-containing metabolite standards, 167 metabolites were positively identified based on retention time and accurate mass matches, which was about 2.5 times the number metabolites identified by 1D-RPLC-MS analysis of the same urine sample.