The purpose of this study was to determine the effect of serum lipids, lipoprotein fractions, and apolipoprotein (apo) A-1, B and E on mortality from vascular and nonvascular causes in an unselected elderly population. The random sample of 347 community-living individuals aged 65 years or older was obtained in 1982. Serum total cholesterol, LDL cholesterol (LDL-C), HDL cholesterol (HDL-C), triglyceride, and apo A-1, B and E were determined at baseline. After the 11-year follow-up, 199 of the participants had died, and 148 were still alive. Mortality data from vascular and nonvascular causes by the end of 1993 were obtained from official registers. In the univariate analysis, a low total cholesterol level was associated with death due to both vascular and nonvascular causes (P value for trend, .021 and .0027, respectively). After the adjustment for other risk factors, the inverse association between total cholesterol and vascular mortality disappeared, but low total cholesterol was still a significant predictor of death due to nonvascular causes. Adjusted relative risks (RRs) of death due to nonvascular causes for those with elevated total cholesterol (5.1 to 6.5, 6.6 to 8.0, and > 8.0 mmol/L) compared with the reference group (< or = 5.0 mmol/L) were 0.5 (95% confidence interval [CI], 0.2 to 1.2), 0.6 (0.2 to 1.0), and 0.2 (0 to 0.8), respectively. Neither concentrations of HDL-C, LDL-C, triglyceride, nor apo B were associated with vascular or nonvascular mortality. On the other hand, low concentration of apo A-1 predicted vascular death. The RR for the lowest tertile was 1.6 (1.1 to 2.5) compared with the highest tertile. Furthermore, the occurrence of the apo E e4 allele was associated with increased risk of vascular mortality (RR, 1.5; 95% CI, 1.0 to 2.2), but the risk was not related to the levels of lipids, lipoproteins, or other apolipoproteins at baseline. Nonvascular mortality also tended to be predicted by the presence of the e4 allele (RR, 1.5; 95% CI, 0.9 to 2.5). In an unselected elderly population, the allelic variation of apo E, i.e., the presence of the e4 allele, and a low concentration of apo A-1 were more accurate indicators of vascular mortality than total cholesterol or lipoprotein fractions. The risk associated with the apo E polymorphism is unrelated to dyslipidemia.