It is not known whether insulin levels, in the setting of insulin treatment, are an independent risk factor for coronary heart disease (CHD). We studied a cohort of 116 insulin-treated individuals, 65 yr or older, who were followed for 5.6-9 yr. All were free of CHD at baseline. There were 47 incident CHD events. In Cox proportional hazards modeling, with fasting immune-reactive insulin levels as a continuous variable, the hazard ratio for CHD was statistically significant (P < 0.0001). When insulin levels were divided into intervals, those in the third interval [43-150 microU/ml (258-900 pmol/liter)] had an adjusted 30% increased relative risk (95% confidence interval, 0.57, 2.98) compared with those in the first interval [<20 microU/ml (<120 pmol/liter)]. Those in the fourth interval [151-400 microU/ml (906-2400 pmol/liter)] had an adjusted 5.6-fold increased risk (2.3-13.1; P < 0.0001). Approximately 15% of the cohort had such elevated insulin levels. Immune-reactive insulin levels were strongly correlated with specific insulin, proinsulin, and insulin antibody levels. Markedly elevated fasting immune-reactive insulin levels were an independent risk factor for CHD in this study of insulin-treated older adults. These observational findings should be confirmed through larger prospective studies, given their implications for insulin therapy.