We analyzed the risk of valve re-replacement in 640 patients reoperated on between 1980 and 1992. This represented 17% of total valve operations (640/3,764) during that period. A univariate and logistic multivariate analysis was carried out for four sequential periods for the 640 re-replacement patients to determine if changing methods of perfusion and myocardial protection affected recent results. There were 323 female and 317 male patients with a mean age of 58 years (range, 17 to 84 years). Ninety-seven (15%) had coronary artery bypass grafting, 135 (21%) were 70 years old or older, 377 (59%) were in New York Heart Association functional class III or less, and 263 (41%) were in functional class IV. The aortic valve was re-replaced in 245, the mitral valve in 289, and both aortic and mitral synchronously in 106. Four periods were analyzed: 1980 through 1982, 1983 through 1985, 1986 through 1988, and 1989 through 1992. The overall operative mortality was 65 of 640 patients (10%), falling from 12/73 (16%) in 1980 through 1982 to 23/268 (8%) in 1989 through 1992 (p = 0.05). Univariate and multivariate logistic analysis documented that New York Heart Association functional class was highly significant for operative mortality; operative mortality was 4% for functional classes I through III, and 19% for functional class IV (p < or = 0.001). The requirement for coronary bypass was of borderline significance (p = 0.05), and year of operation was also significant. Mortality for re-replacement of aortic valve fell from 15% to 10%, double valve from 20% to 9%, and mitral valve from 16% to 6%.(ABSTRACT TRUNCATED AT 250 WORDS)