Little is known about the metabolome fingerprint of pulse consumption. The study of robust and accurate biomarkers for pulse dietary assessment has great value for nutritional epidemiology regarding health benefits and their mechanisms. To characterize the fingerprinting of dietary pulses (chickpeas, lentils and beans), spot urine samples from a subcohort from the PREDIMED study were stratified, using a validated food frequency questionnaire. Non-pulse consumers (≤ 4 g/day of pulse intake) and habitual pulse consumers (≥ 25 g/day of pulse intake) were analysed using a 1H-NMR metabolomics approach combined with multi- and univariate data analysis. Pulse consumption showed differences through 16 metabolites coming from (i) choline metabolism, (ii) protein-related compounds, and (iii) energy metabolism (including lower urinary glucose). Stepwise logistic regression analysis was applied to design a combined model of pulse exposure, which resulted in glutamine, dimethylamine and 3-methylhistidine. This model was evaluated by receiver operating characteristic curve (AUC > 90% in both training and validation sets). The application of NMR-based metabolomics to pulse exposure highlighted new candidates for biomarkers of pulse consumption, the role of choline metabolism and the impact on energy metabolism, generating new hypotheses on energy modulation. Further intervention studies will confirm these findings.