NMR-based metabonomics has been widely employed to understand the stressor-induced perturbations to mammalian metabolism. However, inter-sample chemical shift variations for metabolites remain an outstanding problem for effective data mining. In this work, we systematically investigated the effects of pH and ionic strength on the chemical shifts for a mixture of 9 urinary metabolites. We found that the chemical shifts were decreased with the rise of pH but increased with the increase of ionic strength, which probably resulted from the pH- and ionic strength-induced alteration to the ionization equilibrium for the function groups. We also found that the chemical shift variations for most metabolites were reduced to less than 0.004 ppm when the pH was 7.1-7.7 and the salt concentration was less than 0.15 M. Based on subsequent optimization to minimize chemical shift variation, sample dilution and maximize the signal-to-noise ratio, we proposed a new buffer system consisting of K(2)HPO(4) and NaH(2)PO(4) (pH 7.4, 1.5 M) with buffer-urine volume ratio of 1 : 10 for human urinary metabonomic studies; we suggest that the chemical shifts for the proton signals of citrate and aromatic signals of histidine be corrected prior to multivariate data analysis especially when high resolution data were employed. Based on these, an optimized sample preparation method has been developed for NMR-based urinary metabonomic studies.