To compare the efficacy of Carpentier's tricuspid annuloplasty with De Vega's, we prospectively randomized 159 patients, operated upon between January, 1977, and January, 1980, to one of the two techniques: 76 patients were assigned to the Carpentier group and 83 to the De Vega group. The criterion for inclusion in the study was the presence of moderate to severe tricuspid regurgitation. There were no significant differences in mean age, male proportion, type of mitral lesion, incidence of aortic valvulopathy, and other preoperative and perioperative characteristics between the two groups. However, organic tricuspid damage on macroscopic intraoperative examination was more common in the Carpentier group. At the end of follow-up (average 64 months) in patients with satisfactory left heart hemodynamics, there was a significant difference in the incidence of moderate or severe tricuspid insufficiency between the two groups (De Vega, 14/41; Carpentier, 4/40; p less than 0.01). In 76 patients (40 with Carpentier's annuloplasty and 36 with De Vega's technique), contrast right ventriculography was performed postoperatively and the degree of tricuspid regurgitation assessed semiquantitatively. In both groups, control of tricuspid regurgitation was poor in patients with either high total pulmonary resistances or organic tricuspid damage. If patients with these characteristics are excluded, then significant tricuspid regurgitation was encountered in only one patient in the Carpentier group but in nine of 19 in the De Vega group (p less than 0.01). In conclusion, in the treatment of tricuspid regurgitation, better results are obtained with Carpentier's than with De Vega's annuloplasty, especially if there is no organic tricuspid damage and pulmonary resistances decrease postoperatively.