Abstract Background evidence suggests a reciprocal relationship between cognitive function (CF) and oral health (OH), but no study has demonstrated this inter-relationship in a longitudinal population. Objective to investigate the bidirectional relationship between CF and OH in an ageing cohort. Design cohort study. Setting general community. Subjects participants from the English Longitudinal Study of Ageing. Methods OH, measured by teeth status, self-reported OH and OH-related quality of life (OHRQoL), and CFs were collected at three time points in 2006/07, 2010/11 and 2014/15. Cross-lagged structural equation models were used to investigate the association between CF and OH, adjusted for potential confounding factors. Results 5477 individuals (56.4% women) were included (mean age = 63.1 years at 2006/07, 67.2 at 2010/11 and 70.4 at 2014/15, SD = 8.9) in analyses. The average CF score was 46.5(SD = 12.3) at baseline and 41.2 (SD = 13.4) at follow-up. 3350 (61.2%) participants had natural teeth only and 622 (11.2%) were edentulous. In the fully adjusted model, better cognition at baseline was associated with better OH at follow-up (beta coefficient = 0.02, 95% CI: 0.01–0.03); conversely better OH at baseline predicted better cognition (beta coefficient = 0.12, 95% CI: 0.06–0.18). Similar magnitude and direction of the reciprocal association was evident between cognition and OHRQoL. Conclusions This is the first longitudinal study to demonstrate the positive reciprocal association between CF and OH. The findings suggest the importance of maintaining both good CF and OH in old age.