Community-based health insurance in Lao People's Democratic Republic targets the informal workforce. Estimates of the program's impact on utilization and out-of-pocket expenditures (OOPs) were obtained using a case-comparison study of 3000 households (14 804 individuals) in urban and semi-urban areas. We used propensity score matching to control for bias on observables and to account for heterogeneity. We check the sensitivity of the results using a weighted regression combined with propensity score matching, which leads to doubly robust treatment effect estimates. The results are robust across the two approaches and show that the insured have significantly higher utilization, lower OOPs and lower incidence of catastrophic expenditures, and are less likely to employ coping mechanisms. However, coverage of the scheme is extremely low, indicating negligible population level impact. Furthermore, the results show that the scheme provides greater protection to the better off than to the poor: the poor are less likely to enrol, and among the poor who are enrolled, there has been no significant impact on utilization of outpatient services, total OOPs or catastrophic expenditures. We discuss the policy implications in the context of the international debate regarding the prospects for the role of community-based health insurance in national financing strategies. Copyright © 2013 John Wiley & Sons, Ltd.