During a population survey in 1982-1983 among all community-dwelling elderly aged 65 years and over in East Boston, Massachusetts, measurements of peak expiratory flow rate using the mini-Wright peak flow meter were made on 3,582 participants (80% of those eligible). The average peak flow rate was 315 liters/minute, and a measure of peak flow rate adjusted for age, sex, height, and weight was computed. This was a highly significant (p less than 0.0001) predictor of 5-year total mortality, whose ascertainment was virtually complete. The relative risk was 1.27 (95 percent confidence interval 1.19-1.36) per 100 liters/minute decrease in peak flow rate, using a proportional hazards model including terms for age, sex, and smoking. There was no apparent modification of the effect of this measure in various categories of smoking, with relative risks of 1.24 for nonsmokers, 1.29 for ex-smokers, and 1.26 for current smokers. This finding also persisted after adjustments for other covariates, including respiratory symptoms such as cough, phlegm, and wheeze; cardiovascular risk factors such as history of myocardial infarction and stroke; and systolic and diastolic blood pressures; socioeconomic status; scores on simple tests of cognitive function; measures of physical activity and functional ability; and self-assessed state of health. In a stepwise model including all of these variables, the relative risk was 1.16 (p less than 0.0001) per 100 liters/minute decrease in peak flow rate, indicating that peak flow rate is a strong independent predictor of total mortality in the elderly.