1267 patients who underwent surgical treatment for bronchial carcinoma during the period 1949--70 have been studied. 304 underwent thoracotomy without further resection; 88% died within 12 months; the maximum survival was 56 months. Of the remaining 963 patients, 869 late deaths are recorded, and the causes of these have been studied. The 5-year actuarial survival rate was 25.9%; the 10-year rate 15.3%; the 15-year rate 9.3%; and the 20-year rate 3.9%. 610 of the 869 deaths were due to bronchial carcinoma; 79 were due to respiratory disease; 57 to coronary disease; and 23 to other neoplasms. The percentage of total deaths due to bronchial carcinoma declined with the passage of time. It was 86.5% during the first year; 79.2% during the next 4 years; 53.7% from 5 to 10 years; and 39.3% over 10 years. The second commonest cause of death was respiratory disease, which tended to be related to the extent of resection. Compared with a population of the same age and sex structure, suffering the death rates of England and Wales during the period involved, these patients show a massive excess of deaths from bronchial carcinoma and a smaller excess from respiratory disease; and a considerable shortfall from coronary disease, other carcinomas, and all other causes. A more favourable probability of survival was linked with lobectomy, squamous cell type and the younger age group. A less favourable probability was linked with pneumonectomy, adenocarcinoma and older age-groups. There was no significant difference due to sex. After the first 2 years, with their heavy death rate, the chances of surviving to 10 years improved by about 15% each year.