Recently, Walters ("Dementia: Continuum or distinct entity?", Psychology and Aging, 2010, 25, 534-544) published a taxometric study suggesting a dimensional latent structure for the construct of dementia. However, because that study did not conceptualize dementia according to accepted conventions (i.e., there were no measures of cognitive change or independent functioning), its results may represent a false negative error caused by insufficient content coverage. We replicated Walters, and we used the same taxometric methods and the same data source--but with indicators of cognitive change and functional independence. Our results support a categorical interpretation of dementia; whereas Walters' results suggest that cognitive ability, rather than dementia, is dimensional in nature.