Two alkylresorcinol (AR) metabolites, 3, 5-dihydroxybenzoic acid (DHBA) and 3-(3,5-dihydroxyphenyl)-1-propanoic acid (DHPPA), in urine have been suggested as biomarkers of whole grain (WG) and cereal fiber intake but the long-term reproducibility and correlation with habitual intake has not been determined. Therefore, we evaluated the long-term reproducibility of AR metabolites in spot urine samples and investigated their correlation with habitual WG and cereal fiber intake in U.S. women. AR metabolites were analyzed in 104 women participating in the Nurses' Health Study II and WG and fiber intakes were assessed using a FFQ. Long-term reproducibility was assessed by calculating the intra-class correlation coefficients (ICC) using samples taken 1-3 y (mean 1.8 y) apart. The observed Spearman correlation coefficients (r(s)) and r(s) adjusted for within-participant variation in the biomarker were calculated between WG and fiber intake and biomarkers. The long-term reproducibility was poor for DHBA [ICC = 0.17 (95% CI: 0.05, 0.43)] and modest for DHPPA [ICC = 0.31 (95% CI: 0.17, 0.51)]. The correlation between WG intake in 1995 and DHPPA measured 2 y later was 0.37 (P < 0.0001); the adjusted correlation was 0.60 (95% CI: 0.37, 0.76). Cereal fiber and WG intake were similarly correlated to the biomarkers. DHPPA in spot urine samples reflected WG intake despite relatively low intake of food sources of AR. The poor to modest reproducibility may limit the use of single measurements of these biomarkers in cohort studies in the US, where WG intake is relatively low and has changed over time. But DHPPA in repeated samples may be useful for validating WG intake and assessing compliance in WG intervention studies.