Over one half the patients with rheumatoid arthritis (RA) are being treated with methotrexate (MTX). Although well proven, the efficacy of MTX varies in individual patients. This study examined the metabolic biomarkers that can be used to predict the therapeutic effect of MTX by using metabolomic analysis. Rats were immunized with collagen to rapidly cause collagen-induced arthritis (CIA) and then treated with 0.1 mg/kg MTX for 4 weeks. The clinical signs and the histopathological features of CIA were observed to evaluate the therapeutic effects. Urine samples of CIA rats were collected, and analyzed by using 600 M H-1-nuclear magnetic resonance (H-1-NMR) for spectral binning after the therapy. The urine spectra were divided into spectral bins, and 20 endogenous metabolites were assigned by Chenomx Suite. Multivariate analyses were performed to identify the spectral pattern of endogenous metabolites related to MTX therapy. The results showed that the clustering of the spectra of the urine samples from the responsive rats (n=20) was different from that from the non-responsive rats (n=11). Multivariate analysis showed difference in metabolic profiles between the responsive and non-responsive rats by using partial least squares-discrimination analysis (PLS-DA) (R-2=0.812, Q(2)=0.604). In targeted profiling, 13 endogenous metabolites (uric acid, taurine, histidine, methionine, glycine, etc.) were selected as putative biomarkers for predicting therapeutic response to MTX. It was suggested that H-1-NMR-based metabolomic analysis can be used to predict the therapeutic effect of MTX, and several metabolites were found to be related to the therapeutic effects of MTX.