In the past several decades, individuals 65 and over have experienced remarkable declines in poverty, from 35.2 percent in 1959 to 9.0 percent in 2011. These declines in official poverty rates, however, are based on self-reported income data from the Current Population Survey Annual Social and Economic Supplement (CPS-ASEC). In this paper I evaluate the quality of the retirement income data in the 2010 CPS-ASEC by matching it to individual microdata from IRS 1099-R forms filed with tax returns for the 2009 tax year. I find that recipiency of retirement income may be underreported on the CPS-ASEC, as only a third of individuals with matched 1099-R forms reported receiving any retirement income. Much of this discrepancy, however, may be due to differences in scope, as the 1099-R covers a wider set of payments than is intended to be captured in the CPS-ASEC retirement income items. In cases where positive amounts are reported on the CPS-ASEC, however, those amounts appear to be quite accurate. Among cases with positive reported income, the correlation between CPS-ASEC and 1099-R log amounts is 0.769. Upon modeling 1099-R recipiency as a function of age with discontinuities at various policy-relevant age cutoffs, I find a 14.0 percentage-point jump in recipiency at age 59½, the age at which IRA withdrawals are allowed without penalty. Taken together, these results provide suggestive evidence that retirement income may represent an underestimated contribution to the material well-being of the elderly.