Introduction We sought to assess the strength of associations of smoking with cause-specific mortality in older men.

Methods In a prospective study of surviving participants of the Whitehall study of male civil servants recruited in 1967–1970, 7044 were resurveyed in 1997-1998 (mean age 77 years; range 66–97) and followed for cause-specific mortality. HRs in relation to smoking habits in old age were estimated using Cox proportional hazards models after adjustment for age, prior disease and last known employment grade.

Results At resurvey, 14% of men were current-smokers, 58% were former-smokers and 23% were never-smokers (the remaining 5% reported being a never-smoker in the resurvey, but not in the initial survey). During median follow-up in survivors of 12.5 years, there were 4373 deaths (72/1000/year), of which 1838 were from cardiovascular disease (30/1000/yr), 1044 from cancer (17/1000/yr]); 709 from respiratory (12/1000/yr); and 782 (13/1000/yr) from other causes. Overall mortality rates were higher among current-smokers compared with never-smokers (HR 1.49, 95% CI 1.34 to 1.65), due to increased mortality from cardiovascular disease (1.35, 1.15 to 1.58), cancer (1.71, 1.39 to 2.09), and respiratory disease (2.27, 1.75 to 2.95). Among smokers, mortality rates were linearly related to the number of cigarettes smoked. Former-smokers had only moderately higher overall mortality rates compared with never-smokers (HR 1.16, 1.07 to 1.25), chiefly due to continued increased risks of death from cancer (1.21, 1.03 to 1.42) and respiratory disease (1.60, 1.29 to 1.98).

Conclusions Among men who survived into their 70s, continuing to smoke is associated with persistent excess vascular and non-vascular mortality.