In a Honolulu Heart Program cohort of 1,604 elderly men aged 70-90 years who were sampled from the most recent follow-up examination (1991-1992), current risk factor values were compared with those obtained 25 years earlier when the same men were between ages 45-64 and free of clinically diagnosed cardiovascular disease. Cardiovascular disease risk factors studied included systolic blood pressure, diastolic blood pressure, serum cholesterol, cigarette smoking, body mass index, and alcohol intake. For systolic pressure, 65% of the men had moved into a different quartile by old age, with 25% changing by more than one quartile. For diastolic pressure, 68% had moved into another quartile, with 28% moving more than one quartile, and for body mass index, 53% had moved into another quartile, with 14% moving more than a quartile. Less than 1% started to smoke, while 27% were reclassified from smokers to nonsmokers. Only 4% started to drink alcohol, while 30% were reclassified from drinkers to nondrinkers. When the men were stratified into cardiovascular disease, noncardiovascular disease, and healthy follow-up groups, modest deviations from the overall pattern were observed, with morbidity groups showing a greater tendency to reduction in risk factor levels. The results show that there is a substantial redistribution of major cardiovascular disease risk factor values between midlife and old age. Since midlife values are more likely to represent lifelong exposure values that, in turn, make the main contribution to the development of atherosclerosis, investigators and clinicians may need to be cautious in using risk factor values measured late in life as the only means of assessing risk for subsequent disease.