In this cohort study, 16,911 men who had completed a mailed dietary questionnaire were followed for 11 1/2 years with 721 deaths reported from ischemic heart disease (IHD) (excluding those reporting on the questionnaire a prior history of angina or other heart conditions). Although no association was found between coffee consumption and mortality from IHD, a negative association between coffee consumption and mortality from diseases other than IHD was found. This negative association, found exclusively in the first four years of follow-up, was observed in deaths from digestive diseases, other than malignancies, and paralysis agitans, which made the greatest contribution to this observed negative association. The negative association appeared to reflect a reduction in coffee consumption related to the disorders in question and not to a protective effect of coffee. It has been suggested that the positive association between coffee consumption and IHD reported in some case-control studies may reflect a decreased consumption among controls rather than an unusually high consumption among cases.