ObjectivesThe life course determinants of midlife and later life cognitive function have been studied using longitudinal population-based cohort data, but far less is known about whether the pattern of these pathways is similar or distinct for clinically relevant cognitive state. We investigated this for Addenbrooke’s Cognitive Examination third edition (ACE-III), used in clinical settings to screen for cognitive impairment and dementia.DesignLongitudinal birth cohort study.SettingResidential addresses in England, Wales and Scotland.Participants1762 community-dwelling men and women of European heritage, enrolled since birth in the Medical Research Council (MRC) National Survey of Health and Development (the British 1946 birth cohort).Primary outcomeACE-III.ResultsPath modelling estimated direct and indirect associations between apolipoprotein E (APOE) status, father’s social class, childhood cognition, education, midlife occupational complexity, midlife verbal ability (National Adult Reading Test; NART), and the total ACE-III score. Controlling for sex, there was a direct negative association betweenAPOEε4 and the ACE-III score (β=−0.04 [–0.08 to –0.002], p=0.04), but not betweenAPOEε4 and childhood cognition (β=0.03 [–0.006 to 0.069], p=0.10) or the NART (β=0.0005 [–0.03 to 0.03], p=0.97). The strongest influences on the ACE-III were from childhood cognition (β=0.20 [0.14 to 0.26], p<0.001) and the NART (β=0.35 [0.29 to 0.41], p<0.001); educational attainment and occupational complexity were modestly and independently associated with the ACE-III (β=0.08 [0.03 to 0.14], p=0.002 and β=0.05 [0.01 to 0.10], p=0.02, respectively).ConclusionsThe ACE-III in the general population shows a pattern of life course antecedents that is similar to neuropsychological measures of cognitive function, and may be used to represent normal cognitive ageing as well as a screen for cognitive impairment and dementia.