We propose a lab experiment to understand if and how resource restrictions and uncertainty on the relative needs of future beneficiaries affect physicians’ resource allocation decisions. When there are incentives to over-treat, we find that a patient tended by a constrained physician under uncertainty obtains higher benefits and  receives allocations closer to her optimum than patients from physicians with no constraints or deciding under uncertainty alone. In addition, we observe a redistribution
of resources when physicians decide under resource restrictions and uncertainty. In particular, when resources are scarce, physicians tend to allocate the limited services to patients with higher benefits in the absence of medical services, a higher capacity to benefit from the resources, the scantiest need for service units, and the lowest  benefits at the optimum. Finally, we find that constraints, whether with or without complete information on patient characteristics, lead selfish physicians to approximate to what is best for the patient.