Efficient, valid, and economical methods are needed to measure memory in elderly patients who are participants in clinical trials for the prevention or treatment of dementia. Data provided by knowledgeable informants are an ideal means of assessment, but factors that may limit the validity of informant-report data are not known. This study investigated the living status, relationship type, and educational history of informants and determined the impact of these factors on the validity of informant-report data about patients (N = 62) diagnosed with Alzheimer disease or mild cognitive impairment. Validity of informant-reported memory was indicated by the correlation between the reports and patients' performance on a neuropsychological memory test. Results indicated that informants who lived with patients provided more accurate reports of the patients' memory than informants who did not live with the patient. Spouses were more accurate than other relationship types, although relationship type was confounded with living status. Patient education and neuropsychiatric symptoms were not significantly associated with informant accuracy. Results of this study will aid in selecting informants who can provide the most accurate data about memory disorder patients and will aid in the development of protocols for clinical trials for dementia prevention and treatment.