Relationships between cardiovascular disease (CVD) mortality and breathlessness, a definition of chronic bronchitis, and pulmonary function are investigated among men in two employed populations (17,717 London civil servants and 4904 Scottish workers) and in two communities (844 men in Tecumseh, Michigan and 6859 men in Renfrew and Paisley Burghs, Scotland). Men are aged 40-64 years at entry in all studies except Renfrew-Paisley, where they are aged 45-64 years. Length of follow-up ranges from 6 to 16 years. Age and smoking habits were controlled for in all analyses. Chronic phlegm production is not significantly associated with CVD mortality, and 'chronic bronchitis' is significantly associated with mortality only in the employed populations. Low FEV1 is significantly associated with CVD mortality only in the Whitehall study; however, the rate ratios are above one in all studies. Breathlessness is significantly associated with CVD mortality in all studies. These associations between CVD mortality and 'chronic bronchitis', low FEV1, and breathlessness persist after also controlling for employment grade, systolic blood pressure, antihypertensive medication, ECG changes, plasma cholesterol level, body mass index and diabetes. Only the associations between breathlessness and mortality persist after further controlling for low FEV1 and myocardial ischaemia. The rate ratios between breathlessness and mortality are about two for all studies. It is concluded that in these populations, breathlessness is an independent and major predictor of CVD mortality.