Projection models enable users to assess the costs and benefits of changes to long-term services and supports policies and to compare policy options, using a consistent set of underlying assumptions. Outlined are key challenges model developers face, including data limitations, difficulties anticipating potential behavioral responses, the need to assign appropriate benchmarks, the value of family care and intangibles like autonomy and quality of life, and grappling with uncertainty. The authors also detail substantive lessons learned about policy and describe modeling best practices, including sensitivity analyses.