Nonlinear cost-sharing in health insurance encourages intertemporal substitution because patients can reduce their out-of-pocket costs by concentrating spending in years when they hit the deductible. We test for such intertemporal substitution using data from the RAND Health Insurance Experiment, where people were randomly assigned either to a free care plan or to a cost-sharing plan which had coinsurance up to a maximum dollar expenditure (MDE). Hitting the MDE – leading to an effective price of zero – has a bigger effect on monthly health care spending and utilization than does being in free care, because people who hit the MDE face high future and past prices. As a result, we estimate that sensitivity to short-lasting price changes is about twice as large as sensitivity to long-lasting changes. These findings help reconcile conflicting estimates of the price elasticity of demand for health care, and suggest that high deductible health plans may be less effective than hoped in controlling health care spending.