A study examined the validity of using questions concerning retirement expectations in large-scale micro surveys. A simple static labor supply model that considered wages, hours worked, and health data was developed for this purpose. Next, the model was implemented using data from the National Longitudinal Survey (NLS) on the Labor Market Experience of Older Men. Data from the NLS concerning a sample of white males who were, among other things, (1) between the ages of 45 and 49 in 1966, (2) not self-employed in any survey year, and (3) not subject to mandatory retirement, were compared with actual labor supply data. The reported expected retirement ages were actually more accurate than expected retirement ages predicted by the labor supply model, although the correlation between them is significant. This was found to be consistent with the idea that expectations data provide valuable information about labor supply behavior, which can be exploited to improve estimates of labor supply models. (MN)