To help workers navigate the transition from work to retirement more effectively, employers have been launching phased retirement programs, which allow older employees to work part-time and receive full retirement benefits. This paper examines the experience of the phased retirement system for tenured faculty in the University of North Carolina system over the years 1996–98. After phased retirement was introduced, there was a sizable increase in the overall separation rate in the system. The key finding from an empirical analysis of the retirement decision as a function of pension incentives, employee performance, demographics, and campus characteristics is that the odds of entering phased retirement were strongly and inversely related to employee performance, as measured by recent pay increases.