The Paris Prospective Study I (7434 men) is a long-term investigation of cardiovascular diseases. A previous analysis has shown that high plasma insulin level was a more sensitive marker than glucose intolerance in the prediction of coronary heart disease (CHD). In order to ascertain the risk model for CHD in relation to high circulating plasma insulin level, we studied CHD mortality rates in groups of participants to the study with similar profiles of risk factors. The risk factors considered to cross-classify the subjects were: serum cholesterol level, blood pressure, fasting plasma insulin level (all the significant predictors of CHD mortality in the Cox regression model), plus body mass index (BMI) and 2-h post-load blood glucose level. Serum cholesterol level was linearly related to CHD mortality risk. The strength of the association of blood pressure to the risk was reduced when BMI increased. By contrast, the association of blood pressure to the risk remained linear when plasma insulin level rather than BMI was considered. Plasma insulin level was a more sensitive marker of CHD risk than glucose intolerance in the overweight group. Moreover, in this group, the relation to CHD mortality risk was stronger for plasma insulin level than for blood pressure.