Mean estimates of time preference rates for health vary widely in the literature. If these studies are to inform discounting practice and analyses of health-affecting behaviour, it is crucial to understand why this is the case. One reason for the variation in time preference rates is the use of different elicitation methods. The influence of elicitation method has received little attention in the time preference literature. This study compares directly an open-ended and a closed-ended method. Both private and social time preferences for health are elicited. The closed-ended method produced much lower mean rates than the open-ended method. This is in contrast to the contingent valuation literature which shows that closed-ended methods produce higher estimates of willingness to pay than openended methods. That the elicitation methods produce different mean estimates is clearly worrying if the interest is in estimating the true time preference rate. However, the results of this study suggest that if the interest is in testing different types of time preferences or investigating the relationship between time preference and individual characteristics then the choice of elicitation method is less important. (C) 2008 Elsevier Ltd. All rights reserved.