Aortic valve disease in the elderly is primarily calcific stenosis with preservation of left ventricular function. In contrast, mitral valve disease in the elderly often is ischemic in nature with damage occurring to both valve and myocardium. The present study was undertaken to compare results of aortic (AVR) and mitral valve replacement (MVR) in the elderly and to ascertain predictors of poor outcome. Because patients who had concomitant coronary artery bypass grafting (CABG) are included (51% for AVR, 55% for MVR), patients who had isolated CABG were used as a comparison group. Between January 1, 1984, and June 30, 1991, 1,386 patients aged 70 years and older underwent CABG (n = 1,043), AVR (n = 245), or MVR (n = 98). The operative mortality rates were 5.3% for AVR, 20.4% for MVR, and 5.8% for CABG. Late follow-up of patients undergoing operation in 1984 and 1985 was available for 98% (231/237). Overall survival was comparable for all three groups through the first 5 years of follow-up (AVR, 68% +/- 8%; MVR, 73% +/- 8%; CABG, 78% +/- 3%). After 5 years, survival for patients having AVR and MVR was less than that for those having CABG. Patient age, sex, New York Heart Association functional class, concomitant CABG, prosthetic valve type, native valve pathology, and preoperative catheterization data were examined as possible predictors of outcome by multivariate logistic regression.(ABSTRACT TRUNCATED AT 250 WORDS)