The relation of mortality to ventilatory function and some respiratory symptoms detected during interview has been studied in 3047 men and women followed-up for 13 years. The analysis was performed for all natural causes of death, and separately for deaths due to circulatory diseases and neoplasms. The results confirm the strong predictive power of ventilatory impairment for overall and circulatory mortality, even after adjustment for age, cigarette smoking and other factors in logistic regression models. The risk of cancer death in men increased with decrease of FEV1. However, chronic cough, mucus hypersecretion or asthmatic syndrome were not related to subsequent mortality. A strong predictor of overall and circulatory 'mortality' in men was also self-assessed health status. This was independent of other factors.