BACKGROUND While age-related cognitive decline may affect all stages in the response process--comprehension, retrieval, judgment, response selection, and response reporting--the associations between objective cognitive tests and the agreement between self-reports and Medicare claims has not been assessed. We evaluate those associations using the Survey on Assets and Health Dynamics among the Oldest Old (AHEAD). METHODS Eight waves of reinterviews (1995-2010) were linked to Medicare claims for 3661 self-respondents yielding 12,313 person-period observations. Cognitive function was measured by 2 episodic memory tests (immediate and delayed recall of 10 words) and 1 mental status test (backward counting, dates, and names). Survey reports on 12 diseases and 4 health services were mapped to Medicare claims to derive counts of concordant reports, underreports, and overreports, as were the numbers of hospital episodes and physician visits. GEE negative binomial and logistic regression models were used. RESULTS Better mental status was associated with more concordant reporting and less underreporting on disease history and the number of hospital episodes. Better mental status and delayed word recall were associated with more concordant reporting and less underreporting on health services use. Better delayed recall was significantly associated with less underreporting on the number of physician visits. These associations were not appreciably altered by adjustment for demographic characteristics, socioeconomic status, self-rated health, or secular trends. CONCLUSION We recommend that future surveys of older adults include an objective measure of mental status (rather than memory), especially when those survey reports cannot be verified by access to Medicare claims or chart review.