A retrospective study was undertaken of 31 patients submitted to tricuspid valve replacement between 1977 and 1988, either alone or associated to mitral or aortic replacement. The cause of the tricuspid lesion was rheumatic in 19 patients, congenital in seven, prosthetic valve thrombosis in two and miscellaneous in the remaining 3 patients (endocarditis, carcinoid syndrome, iatrogenic operative tricuspid regurgitation). Of the 31 tricuspid prostheses reviewed, 26 were mechanical (St Jude, Björk-Shiley) and five of the biological type (Ionescu, Hancock). Results showed that 25% of patients died in the hospital; 75% of them had been operated previously. With a follow-up averaging 3.9 years the complications related with the prosthesis were: thrombosis of the tricuspid prostheses in 1 patient (1.2% patients-year) and hemolytic anemia in another. Ninety percent of patient included in the follow-up were in NYHA class I or II. Three patients (14%) died due to a cause not directly related to the prosthesis. An actuarial analysis showed intrahospital survival rates of 55%, 100% and 75% in patient with 3, 2 and 1 prostheses, respectively; at 6 months postoperatively the figures were 55%, 87% and 75%, and at a follow-up of 3 years, 55%, 67% and 66%. In conclusion, in patients submitted to a tricuspid valve replacement mortality is high before hospital discharge, and is related to the clinical situation. A sustained improvement is observed in survivors once discharged, independently of the number of prostheses implanted. Complications related to the tricuspid prosthesis are comparable to those observed with prostheses in mitral or aortic position.