In the U.S. Social Security system, the decision of when to claim Social Security benefits is legally independent of when the individual chooses to separate from the workforce. But if an individual claims benefits prior to his “full retirement age” (FRA) while continuing to have labor earnings above a relatively low threshold, his benefits are reduced via the Social Security Earnings Test. The individual is compensated for this benefit reduction in the form of higher benefit payments payable from the FRA for the remainder of the beneficiary’s lifetime. To the extent that the relevant actuarial adjustment is actuarially fair, the Earnings Test simply represents a re-timing of benefit payments. Nevertheless, many people view the benefit reduction as a tax on earned income after claiming benefits. We posit that whether the Earnings Test influences work and benefit claiming patterns will depend on whether people are aware of the benefit enhancements paid in return for continued work. Using an experimental module of the RAND American Life Panel, we explore how people perceive the Social Security Earnings Test and examine alternative ways to frame the tradeoff between reduced benefits in the short run and higher benefits paid later and for life. Our overall finding is that knowledge of the Earnings Test is uneven, with better educated, higher earning, older individuals showing somewhat greater knowledge than others. The frames we have tested produce only minor effects on individual choices of earnings or claiming ages, and the effects are difficult to reconcile with economic theory.