In a study of 1033 consecutive patients with acute myocardial infarction, serum potassium concentrations were determined on admission to hospital and studied with respect to the subsequent occurrence of atrial fibrillation and flutter and of ventricular tachycardia and fibrillation. The study cohort fulfilled the inclusion criteria for the Norwegian timolol trial in which they later took part. In multivariate analysis, with serum potassium concentrations as a continuous variable, age, the presence of ventricular tachycardia and fibrillation, and maximum level of aspartate aminotransferase greater than four times the upper limit of normal were significantly associated with the occurrence of atrial fibrillation and flutter, while serum potassium concentration was not. Serum potassium concentrations and time from onset of the infarction to hospital admission were significantly negatively associated with the occurrence of ventricular tachycardia and fibrillation; while age, cardiomegaly, transient hypotension, pathological Q-waves in the electrocardiogram, atrial fibrillation and flutter, and ventricular premature beats were positively related to these arrhythmias. Thus, there is an independent inverse relationship between serum potassium concentrations and ventricular arrhythmias in acute myocardial infarction.