Recent studies have led to controversy about whether long-term digoxin therapy after confirmed or suspected myocardial infarction increases mortality. We analyzed the mortality experience in 903 patients enrolled in the Multicenter Investigation of Limitation of Infarct Size (MILIS). As in previous studies, the decision to treat or not to treat with digoxin was made by the patient's personal physician on the basis of the usual clinical indications. Cumulative mortality was 28 percent for the 281 digoxin-treated patients as compared with 11 percent for the 622 patients who did not receive digoxin (P less than 0.001; follow-up interval, six days to 36 months; mean, 25.1 months). However, patients treated with digoxin had more base-line characteristics predictive of mortality than did their counterparts. Adjustment for these differences with two separate applications of the Cox method yielded P values of 0.14 and 0.34 for tests of difference in mortality, providing no evidence for a significant excess mortality associated with digoxin. Thus, the findings in the MILIS population do not support the assertion that digoxin therapy is excessively hazardous after infarction, but the existence of an undetected harmful effect can only be excluded with a randomized study. Until the results of such a study are available, we recommend careful consideration of whether any treatment of ventricular dysfunction is actually needed, consideration of alternatives to digoxin therapy, and restriction of digoxin use to the subgroup of patients (with severe chronic congestive failure and a dilated left ventricle) previously shown to have a beneficial clinical response.