This is the second of two reports dealing with the reliability of measurements of 24-hour urinary electrolyte excretion and blood pressure and estimates of electrolyte-blood pressure associations in INTERSALT, an international study of the relations of electrolyte excretion and other factors to blood pressure, involving more than 10,000 persons from 52 centers in 32 countries. The first report (Am J Epidemiol 1994; 139:927-39) described methods of estimating reliability, taking into account age and sex, and provided estimates for several urinary variables, blood pressure, and pulse rate. This second report uses these estimates of reliability and multivariate procedures to correct associations with blood pressure for "regression dilution bias." In these analyses, the estimated mean change in systolic blood pressure associated with a 100-mmol decrease in sodium intake was -3.1 mmHg with adjustment for age, sex, body mass index, and alcohol intake. For potassium, the estimated mean change in systolic pressure associated with a 15-mmol increase in potassium intake was -1.0 mmHg. These estimates are 44%-50% larger than those previously reported in INTERSALT, based on simple corrections for reliability. Furthermore, compared with previously reported INTERSALT values, the estimated mean change in systolic pressure per unit of change in body mass index decreased by about 5% and that for consumption of > or = 300 ml of absolute alcohol per week increased by about 6%. These results further clarify the observed INTERSALT associations of systolic blood pressure with 24-hour urinary sodium and potassium excretion, body mass index, and alcohol intake.