We evaluate potential determinants of enrollment in an early retirement incentive program for non-tenure-track employees at a large university. Using administrative records on the eligible, population of employees not covered by collective bargaining agreements, historical employee count and layoff data by budget units, and public information on unit budgets, we find dips in per-employee finances in a budget unit during the application year and higher recent per employee layoffs were associated with increased probabilities of eligible employee program enrollment. Our results also suggest that, on average, employees whose salaries are lower than we would predict given their personal characteristics and job titles were more likely to enroll in the early retirement program. To the extent that employees' compensation reflect their productivity, as it should under a pay system in which annual salary increases are based on merit, this finding suggests that adverse selection was not a problem with the program. That is, we find no evidence that on average the "most productive" employees took the incentive. Regression of enrolled in SRI (Yes = 1) on socioeconomic characteristics and employment unit dummies is appended. (Contains 4 figures, 8 tables and 15 footnotes.)