ABSTRACT CONTEXT AND OBJECTIVE: Sodium and potassium intake from different food sources is an important issue regarding cardiovascular physiology. Epidemiological assessment of the intake of these electrolytes intake is done through food frequency questionnaires or urinary excretion measurements. Our aim was to compare these methods using a sample of Brazilian civil servants. DESIGN AND SETTING: Cross-sectional baseline evaluation from the Brazilian Longitudinal Study of Adult Health. METHODS: Sodium and potassium intake was obtained using two methods: a semi-quantitative questionnaire including 114 food items; and overnight 12-hour urinary excretion measurement. Sodium and potassium estimates obtained through the questionnaire were adjusted for energy intake using the residual method. Urinary excretion measurements were considered valid if they met three adequacy criteria: collection time, volume and total creatinine excretion. Mean nutrients were estimated, and Spearman correlations were calculated. Sodium and potassium intake was categorized into quintiles, and weighted kappa coefficients and percentage agreement were calculated. The significance level for all tests was 0.05. RESULTS: Data from 15,105 participants were analyzed, and significant differences between mean intakes of sodium (questionnaire: 4.5 ± 1.7 g; urine: 4.2 ± 2.1 g) and potassium (questionnaire: 4.7 ± 1.8 g; urine: 2.4 ± 1 g) were found. Weak agreement was found for sodium (K = 0.18) and potassium (K = 0.16). The percentage disagreement between methods ranged from 41.8 to 44.5%, while exact concordance ranged from 22.1% to 23.9%. CONCLUSIONS: The agreement between the food frequency questionnaire and urinary excretion measurements for assessment of sodium and potassium intakes was modest.