We study a 2008 policy reform in which Medicare revised its hospital payment system to better reflect patients’ severity of illness. We construct a simulated instrument that predicts a hospital’s policy-induced change in reimbursement using pre-reform patients and post-reform rules. The reform led to large persistent changes in Medicare payment rates across hospitals. Hospitals that faced larger gains in Medicare reimbursement increased the volume of Medicare patients they treated. The estimates imply a volume elasticity of approximately unity. To accommodate greater volume, hospitals increased nurse employment, but also lowered length of stay, with ambiguous effects on quality.