We studied the association between different tests of pulmonary function and subsequent mortality. Subjects were drawn from an epidemiology study of chronic obstructive pulmonary disease. Between 1971 and 1976, 2,539 nonpatient adults had tests of forced expiration, diffusing capacity, and single-breath nitrogen washout. By 1981, 115 of those subjects had died, including 3 from known lung disease. In assessing the relationship between lung function and mortality, the following tests of pulmonary function were examined: forced expiratory volume in one second as a percentage of forced vital capacity, forced expiratory volume divided by height cubed, slope of phase III from the single-breath nitrogen test, closing capacity, diffusing capacity for CO, and the ratio of maximal flow at 75% to that at 25% vital capacity. When adjustments for age and smoking were made, slope of phase III was strongly associated with mortality, even more so than tests of forced expiration. There are two possible explanations for this striking relationship between these observed abnormalities of lung function and subsequent overall mortality (which is largely from nonpulmonary disease): the lungs may serve to protect other systems of the body and therefore poor pulmonary function may contribute to a number of diseases leading to death, or lung function may merely reflect existing disorders in other systems of the body, and the observed association between mortality and pulmonary function is a byproduct of nonpulmonary diseases.