A randomised, double blind, placebo controlled, crossover study of digoxin withdrawal and reintroduction was carried out over two periods of eight weeks each after long term treatment. Forty four patients with stable heart failure in sinus rhythm and plasma digoxin concentrations over 0.8 ng/ml were studied. Their progress was assessed by clinical criteria, by haemodynamic measurements (systolic time intervals and echocardiography), and by pharmacological measurements of erythrocytic sodium pump numbers and activity. After withdrawal of digoxin clinical deterioration occurred in only 25% of the patients. Furthermore, in only 9% of cases was digoxin reintroduction thought to be necessary. There was deterioration in only 11% of the patients during digoxin treatment. Deterioration during digoxin withdrawal was accompanied by changes in systolic time intervals, but similar, albeit smaller changes in systolic time intervals also occurred in patients with no deterioration. Deterioration was accompanied by changes in the pharmacological effects of digoxin on the erythrocytes, consistent with a loss of effect, and these changes did not occur in those who did not deteriorate. The occurrence of deterioration could not be predicted by any clinical, haemodynamic, or pharmacological measurements made before withdrawal.