OBJECTIVES: To evaluate the consistency, stability, and accuracy of reports by patients with Alzheimer's disease (AD) and their caregivers regarding the patients' premorbid and current financial abilities. DESIGN: Consistency of reports was assessed within patient/caregiver dyads and within control/control informant dyads. Stability of reports over a 1-month interval was assessed for each group: patients with AD, caregivers, controls, and control informants. Accuracy of each group's reports was evaluated in reference to patients' and controls' performance on a direct psychometric measure of financial capacity. SETTING: University medical center. PARTICIPANTS: Twenty patients with AD and 20 family caregivers; 23 controls and 23 family informants. MEASUREMENTS: The Prior Financial Capacity Form (PFCF) and the Current Financial Capacity Form (CFCF) were used. Parallel versions assessed self-report (patients, controls) and informant report (caregivers, control informants) at two visits 1 month apart. Patients with AD and controls were also administered the Financial Capacity Instrument (FCI), a direct assessment of the same abilities reported on the PFCF and CFCF. RESULTS: Patients with AD reported that they had more-intact current abilities than their caregivers reported. Patients with AD and their caregivers showed lower levels of stability over time on the PFCF and CFCF than did controls and their informants. Half of the patients with AD overestimated their current abilities relative to their FCI performance, whereas caregivers demonstrated both underestimation and overestimation errors. Controls and informants evidenced high levels of consistency, stability, and accuracy in PFCF and CFCF ratings. CONCLUSION: Patients with AD overestimate their financial abilities in comparison with the reports of their family caregivers. Both patients and caregivers' reports of patients' financial abilities showed limited stability and validity. The reliability and accuracy of self- and informant reports of financial abilities may be compromised in the context of dementia and caregiving, underscoring the need for direct assessment methods to augment self- and informant report in assessing functional decline in dementia.