We randomly assigned 230 patients in sinus rhythm with moderately severe heart failure to treatment with digoxin, milrinone, both, or placebo. The effects of each were compared during a 12-week, double-blind trial. Treatment with milrinone or digoxin significantly increased treadmill exercise time as compared with placebo (by 82 and 64 seconds respectively; 95 percent confidence limits, 44 and 123, and 30 and 100). Both treatments reduced the frequency of decompensation from heart failure, from 47 percent with placebo to 34 percent with milrinone (P less than 0.05; 95 percent confidence limits, 22 and 46) and 15 percent with digoxin (P less than 0.01; 95 percent confidence limits, 7 and 26). However, the clinical condition of 20 percent of the patients taking milrinone deteriorated within two weeks after treatment was begun, as compared with only 3 percent of those taking digoxin (P less than 0.05). The left ventricular ejection fraction at rest was not significantly changed by milrinone (+0.2 percent; 95 percent confidence limits, -1.5 and 1.9), but it was increased by digoxin (+1.7 percent; P less than 0.01; 95 percent confidence limits, -0.03 and 3.4) and decreased by placebo (-2.0 percent; 95 percent confidence limits, -3.8 and -0.1). Three-month survival was related inversely to the base-line ejection fraction. Analysis of mortality from all causes according to the intention to treat suggested an adverse effect of milrinone (P = 0.064). After adjustment for an excess of patients with lower ejection fractions randomly assigned to receive milrinone, this trend was not significant (P = 0.26). Increased ventricular arrhythmias occurred more frequently in patients who received milrinone than in those who did not (18 vs. 4 percent; P less than 0.03). We conclude that milrinone significantly increased exercise tolerance and reduced the frequency of worsened heart failure. However, in the population of patients studied, milrinone or the combination of milrinone and digoxin offered no advantage over digoxin alone. Furthermore, our data suggest that milrinone may aggravate ventricular arrhythmias.