Abstract: Measuring change in cognitive status is essential for the diagnosis, prognosis, and treatment of brain dysfunction. Psychological abilities are differentially affected by brain dysfunction severity, as some abilities are more vulnerable to brain dysfunction than others. Neuropsychological assessments can be viewed as a continuum of 'hold' and 'don't hold' tests. 'Hold' tests assess abilities that remain stable in spite of brain dysfunction, while 'don't hold' tests measure skills that are significantly compromised by brain impairment. The present study ranks the Halstead-Reitan Battery (HRB) and the Wide Range Achievement Test (WRAT) subtests based on their ability to discriminate between two levels of severity in an Alzheimer's Disease (AD) sample. Subtest rankings showed pronounced sex differences, suggesting that test performance is influenced by severity and sex. There appears to be a distinct neuropsychological profile associated with AD. Overall, results indicate that the WRAT is clearly a moderate 'don't hold' test implying that the WRAT cannot be a good estimator of premorbid functioning, as it is moderately related to severity. Use of the WRAT to estimate previous abilities in AD patients would be unwise and would likely underestimate premorbid levels. The HRB is a combination of 'hold' and 'don't hold' subtests, which are directly related to the physiology of the disease process.