As part of its Partnership for Pre-K Improvement (PPI) initiative, the Bill & Melinda Gates Foundation sponsored the RAND Corporation to study the cost of high-quality pre-K programming. The RAND study included three states — Oregon, Tennessee, and Washington — that were partnering with the foundation under PPI. The Gates Foundation selected these states for its PPI initiative partly because they had differences in program requirements (e.g., teacher qualifications) that reflected some of the diversity across state and local pre-K investments. Together, these three states are implementing four pre-K systems: Oregon Pre-Kindergarten, Oregon’s Preschool Promise, Tennessee’s Voluntary Pre-K, and Washington state’s Early Childhood Education and Assistance Program. The RAND researchers then selected three other pre-K systems to include in the study, with standards and evidence of effectiveness as high as, if not higher than, those of the PPI states. Those additional systems are Universal Pre-K Boston, Michigan’s Great Start Readiness Program, and Oklahoma’s Early Childhood Four-Year-Old Program. Including these other systems helped shed light on how much higher per-child costs could go under higher-quality standards. The objective of the study was to estimate the cost per child of a high-quality publicly funded pre-K program in the United States at the state or district level, accounting for cost at the provider level and at the system level). The study was also designed to help stakeholders understand the incremental costs associated with key program features of high-quality systems. The cost analysis described in this report drew on data collected from 36 providers in pre-K systems in Michigan, Oregon, and Tennessee, where such data collection was possible before and during the COVID-19 pandemic. The researchers also incorporated a cost analysis that was conducted in a closely related study based on data collected from pre-K providers in Washington’s Early Childhood Education and Assistance Program. Finally, the project employed a cost model to explore how cost per child varies with key program features. (author abstract)