In this paper, we consider several features of teacher-retention policies based on value-added measures of effectiveness under a variety of empirically grounded rules and parameters. We consider the effects of policy design by varying the standard above which satisfactory teachers are expected to perform. We simulate recently adopted policies that remove teachers based on consecutive unsatisfactory performance and compare these to policies that remove teachers based on poor performance on average over a multiyear period. We also consider the precision of the performance measure and the underlying variation in teacher quality on policy effects. Finally, the simulation makes a step forward by incorporating recent empirical findings of a relationship between teacher quality and natural attrition from the profession. Our results indicate that deselection policies based on value-added measures have the potential to improve teacher quality, although understanding the role of policy design, self-selected exits, and the underlying variation in teacher quality is essential for determining policy effects.