Plasma potassium concentration was measured at admission in 1234 patients who presented with acute chest pain. One hundred and ninety five patients were on beta blockers before admission. The potassium concentrations of patients admitted early (within four hours of onset of symptoms) were compared with those admitted later (4-18 hours after onset of symptoms). There was a transient fall in plasma potassium concentrations in patients not pre-treated with beta blockers. This was not seen in patients who had been on beta blockers before admission. Non-selective beta blockers were more effective than cardioselective agents in maintaining concentrations of plasma potassium. These findings suggest a mechanism for the beneficial effects of beta blockers on morbidity and mortality in acute myocardial infarction.