We investigated the relationship between basal cortisol and dehydroepiandrosterone (DHEA) levels and impairment in different cognitive and noncognitive measures and the possible interaction of DHEA with hypercortisolemia in dementia in 27 patients diagnosed with Alzheimer's disease (AD). There were 17 men and 10 women. Patients were mildly to moderately cognitively impaired at the time of the initial cortisol measures. Patients were administered the Alzheimer's Disease Assessment Scale (ADAS) and Folstein Mini-Mental State Examination (MMSE) at approximately 6-month intervals. Cortisol and DHEA were determined using conventional 125I radioimmunoassay procedures. Pearson product-moment correlations among cortisol and DHEA measures and both initial and longitudinal clinical measures were calculated. There was a relationship between baseline 8 a.m. cortisol levels and cognitive function at the initial testing as measured by the ADAS cognitive measure, with higher cortisol levels being associated with a greater level of impairment. We did not document a relationship between cortisol or DHEA levels and noncognitive measures. There was a significant correlation between both the initial MMSE and ADAS cognitive measures and initial DHEA level, with lower DHEA levels unexpectedly being associated with better performance on these measures. The initial DHEA levels did not predict decline in cognitive function over time. These findings bring into question the potential usefulness of DHEA as a therapeutic agent.