Reported coffee consumption during 1954 to 1958 and 1971 to 1973 was used to test for association with cardiovascular disease (CVD) incidence and lipid values in the Framingham cohort. Multivariate analysis was employed, regressing CVD on age, systolic pressure, cigarette use, body mass index, total cholesterol, and coffee intake. In pooled analyses (2648 men with 549 CVD cases and 3566 women with 462 CVD cases) coffee intake was not associated with CVD incidence in either smokers or nonsmokers, irrespective of sex. Similarly, multivariate analyses for individuals with existing cardiovascular disease showed no association between coffee intake and subsequent cardiovascular disease. In men significant negative associations between coffee and total cholesterol, and very-low-density lipoprotein cholesterol were seen, whereas in women positive associations with low-density lipoprotein cholesterol were observed. Although inconsistent effects on the lipid profile were seen, no increase in primary or secondary CVD was seen with coffee drinking.