Background: To investigate and to compare the relation between dementia and cancer with the association between mild cognitive impairment (MCI) and cancer. Methods: A total of 13,207 persons from the Rotterdam Study were followed between 1990 and 2013 for the onset of dementia and cancer (sample 1). Between 2002 and 2005, a subset of 5,181 persons underwent extensive cognitive testing for MCI and subsequently were followed up for cancer until 2013 (sample 2). We used Cox proportional hazard models to determine the association between dementia and cancer, and MCI and cancer. Results: In sample 1, 1,404 patients were diagnosed with dementia, and 2,316 developed cancer (63 among dementia cases). Dementia was associated with a decreased risk of cancer (hazard ratio [HR] 0.53; 95% CI 0.41–0.68). In sample 2, 513 persons were diagnosed with MCI and 670 persons developed cancer (81 among MCI cases). In contrast to individuals with dementia, those with MCI tended to have an increased risk of cancer (HR 1.25; 95% CI 0.99–1.58). Conclusions: We found that persons with MCI tended to have an increased risk of cancer, whereas those with dementia have a decreased risk. These findings call into question a biological explanation for the inverse link between dementia and cancer, thereby suggesting the presence of methodological bias.