Background Physical activity (PA) is inversely associated with mortality in the general population. We wanted to quantify the association of self-reported PA with mortality from all causes, ischaemic heart disease (IHD) and stroke, and compare it with other known risk factors in different age segments.

Methods The Bergen Clinical Blood Pressure Survey examined a sample of 6811 Norwegian men and women in 1965–71 with follow-up until 2005–07. Cox proportional hazard regression ratio (HR) and population attributable fraction (PAF) were calculated for the old (>65), middle-aged (45–64) and young adults (22–44), respectively. We minimized confounding and bias by progressive comprehensive adjustments and subgroup-analyses (excluding early follow-up deaths, participants with self-reported disease and participants with changes in their PA-level prior baseline due to disease).

Results The HR [95% confidence interval (CI)] associated with a high PA-level was 0.63(0.56–0.71), 0.66(0.52–0.83) and 0.66(0.47–0.93) for mortality from all causes, IHD and stroke, respectively (reference: no participation in any of the listed activities, adjusted for age and gender). PAF (95% CI) of no/low activity (reference: any activity) was consistent across all age groups, varying from 7.3% (3.4–11.4) in the young adults to 9.1% (3.6–15.3) in the old. PAF of smoking and high s-cholesterol declined with increasing age [smoking from 19.9% (15.3–24.7) to 1.5% (−1.3 to 6.2) and s-cholesterol from 11.5% (5.6–17.5) to −9.5% (−18.1 to −0.7)], whereas PAF of hypertension increased from 5.3% (2.1–9.1) to 18.9% (8.3–28.4).

Conclusion The relative importance of traditional risk factors varies between the age groups, but physical activity is a major health promoting factor across all age segments and should be encouraged particularly in an ageing population.