NMR-based metabolite profiling of urine is a fast and reproducible method for detection of numerous metabolites with diverse chemical properties. However, signal overlap in the (1)H NMR profiles of human urine may hamper quantification and identification of metabolites. Therefore, a new method has been developed using automated solid-phase extraction (SPE) combined with NMR metabolite profiling. SPE-NMR of urine resulted in three fractions with complementary and reproducible sub-profiles. The sub-profile from the wash fraction (100 % water) contained polar metabolites; that from the first eluted fraction (10 % methanol-90 % water) semi-polar metabolites; and that from the second eluted fraction (100 % methanol) aromatic metabolites. The method was validated by analysis of urine samples collected from a crossover human nutritional intervention trial in which healthy volunteers consumed capsules containing a polyphenol-rich mixture of red wine and grape juice extract (WGM), the same polyphenol mixture dissolved in a soy drink (WGM_Soy), or a placebo (PLA), over a period of five days. Consumption of WGM clearly increased urinary excretion of 4-hydroxyhippuric acid, hippuric acid, 3-hydroxyphenylacetic acid, homovanillic acid, and 3-(3-hydroxyphenyl)-3-hydroxypropionic acid. However, there was no difference between the excreted amounts of these metabolites after consumption of WGM or WGM_Soy, indicating that the soy drink is a suitable carrier for WGM polyphenols. Interestingly, WGM_Soy induced a significant increase in excretion of cis-aconitate compared with WGM and PLA, suggesting a higher demand on the tricarboxylic acid cycle. In conclusion, SPE-NMR metabolite sub-profiling is a reliable and improved method for quantification and identification of metabolites in urine to discover dietary effects and markers of phytochemical exposure.