Public school choice has expanded rapidly across U.S. districts over the past two decades. We use original data to document that most districts operate opt-in choice systems characterized by highly selective participation. To examine how the design of public school choice systems shapes participation and benefits, we conduct a detailed study of the Los Angeles Unified School District (LAUSD) using nearly two decades of administrative data, oversubscribed admission lotteries, and quasi-experimental expansions in school access. Participation is highly selective, consistent with the national evidence, and lottery estimates show that the lowest-demand students would gain the most from attending choice schools—implying voluntary participation screens out high-return students. To move from reduced-form facts to system-wide policy analysis, we estimate a structural model linking application, enrollment, and achievement, identified using policy-driven sources of variation. Choice schools are vertically differentiated and raise achievement on average, but high application costs and negative selection on gains limit their impact. Counterfactual simulations evaluate multiple design levers— reducing information frictions, lowering transportation barriers, or mandating participation—and quantify their effects on sorting, capacity utilization, and achievement. Counterfactual simulations show that transportation and mandate policies would have the largest impacts on unused capacity and substantially raise district-level achievement. These results underscore that system design, not school effectiveness alone, determines who benefits from public school choice and by how much.