Background Independently of cardiovascular disease (CVD) risk factors, cognitive ability may account for some of the excess risk of coronary heart disease (CHD) associated with lower education. We aimed to assess how late adolescence cognitive ability and midlife CVD risk factors are associated with the educational gradient in CHD in Norway.

Methods In a cohort of 57 279 men born during 1949–1959, health survey information was linked to military conscription records of cognitive ability, to national educational data, to hospitalisation records from the Cardiovascular Disease in Norway (CVDNOR) project and to the Norwegian Cause of Death Registry.

Results Age and period adjusted HR for incident CHD events was 3.62 (95% CI 2.50 to 5.24) for basic relative to tertiary education, and was attenuated after adjustment; to 2.86 (1.87 to 4.38) for cognitive ability, to 1.90 (1.30 to 2.78) for CVD risk factors, and to 1.84 (1.20 to 2.83) when adjusting for both. Age and period adjusted absolute rate difference was 51 (33 to 70) incident CHD events per 100 000 person years between basic and tertiary educated, and was attenuated after adjustment; to 42 (22 to 61) for cognitive ability, to 25 (7 to 42) for CVD risk factors, and to 24 (5 to 43) when adjusting for both.

Conclusions Late adolescence cognitive ability attenuated the educational gradient in incident CHD events. CVD risk factors further attenuated the gradient, and to the same extent regardless of whether cognitive ability was included or not. Cognitive ability might be linked to the educational gradient through CVD risk factors.