Agreement between overnight and 24-hour urinary sodium, potassium, calcium, and magnesium excretion was studied in a sample of 63 normotensive Southwestern Chinese men: 30 Yi farmers and 33 urban residents in April 1989. Overnight (8-hour) and 24-hour urine specimens were collected on 3 consecutive days. Estimated correlation coefficients between 24-hour and overnight mean true values were 0.863 and 0.906 for sodium, 0.736 and 0.816 for potassium, 0.902 and 0.725 for calcium, and 0.733 and 0.703 for magnesium in Yi farmers and urban residents, respectively. Hourly overnight urinary sodium and potassium excretion rates were significantly lower than the corresponding hourly 24-hour urinary excretion rates: -0.60 and -1.99 mmol/hour for sodium, -1.24 and -0.48 mmol/hour for potassium (all p < 0.05) in Yi farmers and urban residents, respectively. In multiple regression analyses, the differences between 24-hour and overnight urinary sodium and potassium excretion rates were significantly and positively related to differences between 24-hour and overnight creatinine excretion rates. The ratios of intraindividual to interindividual variance were lower for 24-hour collections than for overnight collections for sodium and calcium, but the differences in these ratios for potassium and magnesium were small. For sodium and calcium, twice as many overnight as 24-hour collections were required to estimate the correlation between cations and blood pressure with the same accuracy; for potassium and magnesium, overnight and 24-hour collections were equally accurate. These results indicate that in normotensive populations such as the one studied, overnight urine collections may be used to estimate 24-hour cation excretion. The underestimate of cation excretion by assessments based on collection of overnight specimens may be due to either a lower creatinine clearance or a lower intake of cations at night.