To assess the association between lipid levels and cardiovascular events in older adults. A prospective population-based study. Four field centers in U.S. communities. A total of 5,201 adults aged 65 and older living in U.S. communities, plus a recruitment of 687 African Americans 3 years later. Fasting lipid measures included low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), total cholesterol, and triglycerides. At baseline, 1,954 men and 2,931 women were at risk for an incident myocardial infarction (MI) or stroke. During an average 7.5-year follow-up, 436 subjects had a coronary event, 332 had an ischemic stroke, 104 a hemorrhagic stroke, and 1,096 died. After adjustment, lipid measures were not major predictors of the outcomes of MI, ischemic stroke, hemorrhagic stroke, and total mortality. For total cholesterol and LDL-C, the associations with MI and ischemic stroke were only marginally significant. HDL-C was inversely associated with MI risk (hazard ratio=0.85 per standard deviation of 15.7 mg/dL, 95% confidence interval=0.76-0.96). For the outcome of ischemic stroke, high levels of HDL-C were associated with a decreased risk in men but not women. Lipid measures were generally only weakly associated with the risks of hemorrhagic stroke or total mortality. In this population-based study of older adults, most lipid measures were weakly associated with cardiovascular events. The association between low HDL-C and increased MI risk was nonetheless strong and consistent.