This paper studies how congestion in the public health sector can be used as both an in‐kind and in‐cash redistributive tool. In our model, agents differ in productivity and they can obtain a health service either from a congested public hospital or from a noncongested private one at a higher price. With pure in‐kind redistribution, agents fail to internalize their impact on congestion, and the demand for the public hospital is higher than optimal. When productivities are not observable but the social planner can assign agents across hospitals, the optimal congestion is higher than in the full information case in order to relax incentive constraints and foster income redistribution. Finally, if agents can freely choose across hospitals, the optimal subsidy on the private hospital price may be negative or positive depending on the relative importance of redistribution and efficiency concerns. In this case, redistribution is limited if the quality of the public facility depends on the number of users.