Abstract We determined gastrin and pepsinogen I in serum samples obtained in 1968-69 from 256 women (175 women 54 years old and 81 women 60 years old when sampled). The concentration of gastrin in serum was significantly (p less than 0.01) and positively correlated with the incidence of myocardial infarction during a 12-year follow-up, with age taken into account as a background variable in multivariate analysis. It was not correlated with overall mortality or with the 12-year incidences of angina pectoris, electrocardiographic changes indicating ischemic heart disease, or stroke. The correlation with myocardial infarction was independent of smoking, systolic blood pressure, indices of obesity, concentrations of blood glucose during fasting and of serum triglycerides and cholesterol, and of the presence of diabetes mellitus at screening or during follow-up. Serum gastrin was significantly (p less than 0.05) related to body mass index (positive age-specific relation) and to smoking (negative age-specific relation). These findings may provide a new aspect to analysis of risk factors for myocardial infarction.