This research estimates the impact of the New Cooperative Medical Scheme (NCMS), a public health insurance program in rural China, on health service use, healthcare costs, and health outcomes. Using difference-in-difference and propensity score matching methods, we address county and household selection bias with panel data from the China Health and Nutrition Survey. By comparing various methods used in prior studies of NCMS to define treatment and comparison groups, we find evidence of county selection bias. Taking a new sampling approach after controlling for county selection issues, we find that while NCMS does not improve health, use of both preventive services and township hospitals increases, as do costs for treating the common cold. In particular, we find that relatively vulnerable households – those with lower incomes and older members participating in NCMS–travel greater distances to access healthcare services, incurring higher travel and treatment costs, yet with no measurable health benefits.