Background The potential association between level of education and age-related cognitive decline remains an open question, partly because of a lack of studies including large subsamples with low education and follow-up intervals covering a substantial part of the adult lifespan. Objectives To examine cognitive decline assessed by a comprehensive clinical test of intelligence over a 35-year period of follow-up from ages 50 to 85 and to analyze the effect of education on trajectories of cognitive decline, including the effects of selective attrition. Methods A longitudinal cohort study with a 35-year follow-up of community dwelling members of the Glostrup 1914 cohort. The study sample comprised 697 men and women at the 50-year baseline assessment and additional participants recruited at later follow-ups. Verbal, Performance, and Full Scale IQs were assessed using the Wechsler Adult Intelligence Scale at ages 50, 60, 70, 80, and 85. To be able to track cognitive changes between successive WAIS assessments, all IQs were based on the Danish 50-year norms. Information on school education was self-reported. The association between education and cognitive decline over time was examined in growth curve models. Selective attrition was investigated in subsamples of participants who dropped out at early or later follow-ups. Results The trajectories for Verbal, Performance, and Full Scale IQ showed higher initial cognitive performance, but also revealed steeper decline among participants with a formal school exam compared to participants without a formal exam. Verbal IQ showed the largest difference in level between the two educational groups, whereas the interaction between education and age was stronger for Performance IQ than for Verbal IQ. In spite of the difference in trajectories, higher mean IQ was observed among participants with a formal school exam compared to those without across all ages, including the 85-year follow-up. Further analyses revealed that early dropout was associated with steeper decline, but that this effect was unrelated to education. Conclusion Comprehensive cognitive assessment over a 35-year period suggests that higher education is associated with steeper decline in IQ, but also higher mean IQ at all follow-ups. These findings are unlikely to reflect regression towards the mean, other characteristics of the employed test battery or associations between educational level and study dropout.