This study aimed to assess the extent to which first-morning void (FMV) urine samples can estimate sodium and potassium excretion compared with 24-hour (24-h) urine samples at the population level. We conducted a cross-sectional study collecting urine samples (FMV and 24-h) and two non-consecutive 24-h dietary recalls in a sub-sample from the Portuguese IAN-AF sampling frame. Six predictive equations were used to estimate 24-h sodium and potassium excretion from FMV urine samples. Pearson correlation coefficients were calculated to compare the association between FMV and 24-h urine collections. Cross-classifications into tertiles were computed to calculate the agreement between measured and estimated excretion with and without calibration. Pearson correlation coefficients were calculated to compare the excretion estimation from FMV and reported intake from 24-h dietary recalls. Bland–Altman plots assessed the agreement between two-day dietary recall and the best-performing calibrated equation. Data from eighty-six subjects aged 18–84 were analysed. Estimated sodium and potassium concentrations from the predictive equations moderate or strongly correlated with the measured 24-h urine samples. The Toft equation was the most predictive and reliable, displaying a moderate correlation (r=0.655) with no risk of over or underestimation of sodium excretion (p=0.096). Tanaka and Kawasaki equations showed a similar moderate correlation (r=0.54 and r=0.58, respectively) but tended to underestimate the 24-h urine excretion of potassium (p<0.001). Calibrated predictive equations using FMV urine samples provide a moderately accurate alternative and resource-efficient option for large-scale nutritional epidemiology studies when 24-h urine collection is impractical.