Following Wave I HRS respondents for six waves (12 years) so that their actual retirement can be observed shows that the actual retirement hazard is substantially higher at (and around) the age that workers identified in Wave I as the “usual” retirement age for workers like them. This is true even when we control for actual age at each wave, and for baseline values of earnings, wealth, health, and marital status. We find relatively consistent evidence that those who report that there is no “usual” retirement age for workers like them tend to retire earlier than other workers – indeed, they are more likely to retire than workers who are more than three years short of their usual retirement age. The finding that workers are more likely to retire at a particular age if they regard that age as the usual retirement age for workers like them suggests that the direct measures of the usual age may be useful in more formal models of the retirement process. In a world where some workers understand the incentives they face and respond appropriately, but others are poorly informed and overwhelmed by the choices they face, the “usual” retirement age may be a starting point for modeling the behavior of the latter group of workers.