The authors present the current status of surgery for the cardiovascular manifestations of the Marfan syndrome. In addition, a brief review of current Marfan genetic research is presented. Data on all Marfan patients undergoing aortic root replacement at the Johns Hopkins Hospital (September 1976-June 1995) were analyzed. Survival and event-free curves were calculated and risk factors for early and late death were determined by univariate and multivariate analysis. Two hundred twelve Marfan patients underwent aortic root replacement using composite graft (202), homograft (8) or valve-sparing procedures (2). One hundred eighty-five patients underwent elective repair with no 30-day mortality. Twenty-seven patients underwent urgent surgery, primarily for acute dissection; two patients with aortic rupture died in the operating room. Actuarial survival of the 212 patients was 88% at 5 years, 78% at 10 years and 71% at 14 years. By multivariate analysis, only poor NYHA class, male gender and urgent surgery emerged as significant independent predictors of early or late mortality. Histologic examination of excised Marfan aortic leaflets by immunofluorescent staining for fibrillin showed fragmentation of elastin-associated microfibrils. These studies suggest cautious use of valve-sparing procedures in Marfan patients. Over the last 5 years significant progress has been made in identifying mutant genes that code for defective fibrillin microfibrils in Marfan patients. Attempts are underway to develop animal models of Marfan disease for study of possible gene therapy. Aortic root replacement can be performed in Marfan patients with operative risk under 5%. Long-term results are gratifying. At present, valve-sparing procedures should be used cautiously in Marfan patients because of fibrillin abnormalities in the preserved aortic valve leaflets.