Transient hypokalaemia may occur in acutely ill patients and is associated with an increased incidence of life-threatening arrhythmias. Therefore, we performed a retrospective analysis of the serum potassium values of 538 patients with unstable angina included in the Holland Interuniversity Nifedipine/metoprolol Trial in relation to the use of diuretics. On admission, 113 of these patients used diuretics. Potassium sparing diuretics had been used in 65 patients (group A) and non-potassium sparing diuretics in 48 patients (group B). From the 425 patients not on diuretics a random sample of 56 (group C) was drawn. Blood samples were taken routinely on admission to the coronary care unit. The serum potassium values found for group A, B and C were 3.77 +/- 0.55, 3.44 +/- 0.69 and 4.14 +/- 0.48, respectively, and the prevalence of hypokalaemia (less than 3.6 mmol) 40, 65 and 14% respectively. Rate ratio [95% confidence interval (CI)] for hypokalaemia when compared to group C was 2.6 (1.2-5.6) for group A and 4.9 (2.4-10.1) for group B. The prevalence of hypokalaemia was higher for women than for men (rate ratio, 95% CI: 1.4, 0.9-2.2). Patients already on beta-blocker therapy showed a 10% lower prevalence of hypokalaemia (rate ratio, 95% CI: 0.7, 0.5-1.1). These data were compared with serum potassium values of 104 patients with stable angina, who reported to the outpatient clinic. These patients were also divided into three groups according to the use of diuretics. Only in 15% of the patients using non-potassium sparing diuretics was hypokalaemia observed.(ABSTRACT TRUNCATED AT 250 WORDS)