Background: To identify associations between polyphenol intake and health and disease outcomes in cohort studies, it is important to identify biomarkers of intake for the various compounds commonly consumed as part of the diet. Objective: The objective of this systematic review was to assess the usefulness of polyphenol metabolites excreted in urine as biomarkers of polyphenol intake in humans. Design: The method included a structured search strategy for polyphenol intervention studies on Ovid MEDLINE, EMBASE (Ovid), and Cochrane databases; formal inclusion and exclusion criteria; data extraction into an Access database; validity assessment; and meta-analysis. Results: One hundred sixty-two controlled intervention studies with polyphenols were included, and mean recovery yield and correlations with the dose ingested were determined for 40 polyphenols. Polyphenols such as daidzein, genistein, glycitein, enterolactone, and hydroxytyrosol showed both a high recovery yield (12-37%) and a high correlation with the dose (Pearson's correlation coefficients: 0.67-0.87), which showed good sensitivity and robustness as biomarkers of intake throughout the different studies. Weaker recovery for anthocyanins (0.06-0.2%) and weaker correlations with dose [Pearson's correlation coefficients: 0.21-0.52 for hesperidin, naringenin, (-)-epicatechin, (-)-epigallocatechin, quercetin, and 3 microbial metabolites of isoflavones (dihydrodaidzein, equol, and O-desmethylangolensin)] suggest that they are currently less suitable as biomarkers of intake. Conclusions: These data confirm the value of certain urinary polyphenols as biomarkers of intake. A validation in populations is now needed to evaluate their specificity, sensitivity, and responsiveness to dose under free-living conditions.