The relationship was examined in 459 male London Civil Servants between the casual lying systolic and diastolic (phase V) blood pressures measured in the clinic, and 24-hr urinary sodium and potassium excretion collected an average of 6 weeks later. Systolic and diastolic pressures, after adjusting for age and weight, were both negatively associated with the 24 hr potassium excretion (r = -0.17 [p less than 0.001, 95% confidence limits -0.07, -0.27) and r = -0.11 (p less than 0.05, 95% CL 0.00, -0.21) for systolic and diastolic pressure respectively]. Adjusted systolic but not diastolic pressure was also negatively associated with plasma potassium estimated at the time of blood pressure measurement (r = -0.15, p less than 0.001, 95% CL -0.06, -0.24). In 159 female London Civil Servants both systolic and diastolic pressures were negatively correlated with plasma potassium. However, there was no evidence for a negative relationship between urinary potassium excretion and blood pressure in women. Urinary sodium excretion was not related to blood pressure either in men or women. Blood pressure was also measured by the subjects at home over a one week period during which time the urine collection was made. There was no evidence for an association between blood pressure measured at home 6 weeks after the time of drawing blood, and plasma sodium. On the other hand, in men, the negative relationship between blood pressure and urinary potassium excretion was consistent and independent of the time and place of measuring blood pressure. This is consistent with a true effect of diet, a low potassium intake being associated with an increased blood pressure.(ABSTRACT TRUNCATED AT 250 WORDS)