Salthouse (2011) argued that (a) variance between individuals on cognitive test scores remains constant between 20 and 90 years of age and (b) widely recognized problems of deducing functional relationships from patterns of correlations between measurements become especially severe for neuropsychological indices, especially for gross indices of age-related brain changes (e.g., losses of brain volume or increases in white matter lesions). I argue that between-individual variability on cognitive tests does increase with age and provides useful information on causes of age-related cognitive decline. I suggest that problems of inference from correlations are just as difficult for behavioral as for neurophysiological indices and that inclusion, in analyses, of even gross measures of brain status such as loss of volume and white matter lesions can correct misinterpretations that occur when only behavioral data are examined.