After dropping for a century, the average retirement age for U.S. males seems to have leveled off in recent decades. An important question is whether as future improvements in technology cause wages to rise, desired retirement ages will resume their downward trend, or not. This paper attempts to use HRS panel data to test how relatively high (or low) earnings affect male retirement ages. Our goal is to use cross—sectional earning differences to help anticipate likely time—series developments in coming decades. Our preliminary regression results show that higher earnings do lead to somewhat earlier retirement. Unless additional analysis changes the parameter estimates, the implication is that the downward trend in male retirement ages will ultimately return.